Institutional Quality, Macroeconomic Policy, and Sustainable Growth in Thailand

Abstract

The effectiveness of fiscal and monetary policy in sustaining growth and facilitating recovery from economic crises is increasingly considered to be significantly influenced by the quality of a country’s institutions. Strong institutions may determine how well macroeconomic policies perform under both stable and turbulent circumstances. This study examines how institutional quality (IQ) moderates the effects of fiscal and monetary policies on economic growth in Thailand from Q1:2003 to Q4:2023. Using a combination of BART and BASAD models, we find that voice and accountability and control of corruption are key institutional factors. Among macroeconomic indicators, exports, household debt, gold prices, and electricity generation emerge as the most important drivers of growth during the study period. The findings showed that IQ stabilizes and enhances the impact of policy interest rates and export growth while mitigating negative shocks from household debt and energy infrastructure challenges. Monetary policy effectiveness varies and depends on governmental institutions. Fiscal policy remains mostly neutral but shifts with institutional conditions. These results highlight that strong institutions improve the efficacy of macroeconomic policies and support sustainable growth. This study empirically examines the moderating role of IQ in economic resilience and policy design in an emerging economy using microdata from Thailand as a focus and the Time-varying Seemingly Unrelated Regression Equation (tvSURE) model.

1. Introduction

Sustained long-term economic growth remains a focal concern for both developed and developing economies, especially in small open and emerging markets, where structural vulnerabilities and external shocks are impacted by the efficacy of macroeconomic policies. Keynes [1] argued that fiscal policy boosts aggregate demand, especially during economic downturns, through increased government spending or tax cuts. Similarly, Friedman and Schwartz [2] suggested that monetary policy influences output by adjusting policy rates and money supply to attract more investment and consumption. Nearly all countries implement fiscal and monetary policies and tools to manage market fluctuations and sustain economic stability. This raises a question: “Why do similar macroeconomic policies not return the same outcomes in different nations?” Yet, these theories assume an institutional environment in which policies are effectively implemented. Clearly, growth is not driven by policy alone. Governments may increase public expenditure or cut rates, but the success of these measures depends on institutions and leadership.

North [3] defined institutions as the formal and informal rules that govern political, economic, and social relations. According to institutional economic theory [3], strong institutions create incentives for higher productivity and inspire public trust. In contrast, weak institutions can weaken public trust and distort incentives. As per the results, it is difficult to attain the expected policy outcomes under weak institutional governance. For example, capital inflows and private investment tend to be discouraged in the absence of fundamental institutional elements such as freedom of expression and rule of law. Thus, strong institutions are essential for economic resilience, as they ensure that fiscal spending reaches its targets and monetary adjustment meets its expectations. In addition, institutions become one of the main driving engines of the triumph of macroeconomic policies. These results highlight that sustainable economic growth can be achieved when policies are implemented under strong institutions by trusted economic agents.

A recently growing amount of literature has supported these arguments, including the work of [4,5,6,7,8,9,10,11,12]. Institutions mediate between policy design and economic outcomes. Strong institutions reduce transaction costs and promote confidence, while weak organizations increase corruption and inefficiency [4]. These authors emphasized the historical roots of institutions and their long-term effects on development. Rodrik, Subramanian and Trebbi [8] found that institutional quality (IQ) is more important for growth than are trade or geography, making institutions central to development strategies. Keefer, Knack and Boondoggles [10] explained that weak institutions lead to poor policy enforcement, harming investor trust and slowing growth. Kirsanova, Stehn and Vines [9] linked fiscal policy performance to institutional settings in transition economies, showing that policy effectiveness depends on good governance structures. Acemoglu and Robinson [5] further developed a theory of inclusive vs. extractive institutions, arguing that growth is only sustainable when institutions promote broad participation and accountability. Bon [6] applied this framework to ASEAN countries and found that only nations with strong institutions benefited from fiscal expansion. Radulović [12] studied Balkan economies and confirmed that IQ moderates the link between monetary policy and inflation control. Similarly, Khan, Raza and Vo [11] showed that the effect of fiscal spending on output is positive only under strong governance. Recently, Kakar, Younas and Malik [7] tested the effect of corruption control on growth and found mixed results in emerging economies. Still, empirical insights remain scattered, static, and overlook short-term and country-specific dynamics. Therefore, this study builds on these findings by empirically testing how IQ shapes the efficacy of macroeconomic policies in Thailand.

This institutional breadth is vastly relevant to the case of Thailand. The country has implemented several fiscal and monetary policies over the last two decades. However, the outcomes remain inconsistent. Thailand has experienced periods of effective economic stimulus followed by events of slow recovery and rising household debt [13,14]. Recent empirical work confirms this uneven performance, showing that while monetary policy tends to be more consistent in sustaining growth, fiscal policy is more effective during recessions and when localized [15]. These mixed effects cannot be explained without looking at institutional factors such as political instability, limited voice, and accountability, as well as corruption. Therefore, empirical analysis of how IQ influences the effectiveness of fiscal and monetary policies on economic growth is essential for explaining the growth dynamics in the Thai economy.

In this study, we use a two-step method to examine how IQ affects the link between macroeconomic policies and economic growth in Thailand, covering 84 quarters between Q1 2003 and Q4 2023. First, we apply Bayesian Additive Regression Trees (BART) and Bayesian Shrinking and Diffusing Priors (BASAD) models to choose the most important variables. These identify which input institutional and macroeconomic factors matter most for growth during the studied periods. Next, we use the Time-Varying Seemingly Unrelated Regression Equations (tvSURE) model to empirically explore how the effects of fiscal and monetary policies shift over time, with and without the inclusion of selected IQ indicators. This approach examines whether strong institutions improve the effectiveness of fiscal and monetary policies in sustaining growth. To guide our analysis, we focus on the following testable claims: (1) the effects of fiscal and monetary policies on economic growth are time-varying, and (2) IQ impacts the effectiveness of these macroeconomic policies in Thailand.

This study is organized as follows: Section 1 presents introduction; Section 2 reviews relevant studies; Section 3 specifies research methods; Section 4 provides data used and discusses the empirical findings; and finally, Section 5 concludes the study with potential implications and limitations.

2. Review of Relevant Studies

Institutions perform as a mediator between policy design and economic outcomes; consequently, sustainable growth depends not only on policy choices but also on the strength of the institutions that support them. Weak institutions increase corruption and inefficiency, while strong institutions cut transaction costs and lower uncertainty [16]. In developing countries, these effects have been obvious. In developing economies, FDI has positively correlated with economic growth only when IQ has exceeded a certain threshold [17]. They also concluded that institutional reforms have been necessary to unlock the growth benefits of FDI.

Empirically, Nguyen et al. [18] has shown that IQ has had a significant positive effect on economic growth in 29 emerging economies during 2002–2015. Similarly, Nayak and Pradhan [19] has found that governance indicators have significantly boosted GDP, GDP per capita, and the Human Development Index (HDI) across 47 Asian countries between 1981 and 2021. Lopes et al. [20] has reported that regulatory quality has positively influenced growth in BRICS countries, although they have observed a negative effect of the rule of law on growth from 1996 to 2018. Besides, Ramadhan [21] has also confirmed that IQ and governance performance have driven economic growth in Indonesia. Christiano, Eichenbaum and Evans [22] has also concluded that the importance role of exogenous monetary policy shocks in shaping institutions and rules.

Fiscal policy influences long-term economic growth primarily through public spending and taxation. Keynes [1] argued that government expenditure can stimulate demand during economic downturns. Marioli, Fatas and Vasishtha [23] has supported this view but noted that fiscal policy is more volatile in emerging economies than in advanced ones. This reduces growth and clarifies part of the income gap across countries. However, Barro [24] has warned that excessive spending may force out private investment and potentially slow growth. Few empirical studies have been consistent with what [24] concerned, emphasizing that the efficiency of government spending matters more than its size [11,25,26]. In Thailand, public investment in electricity and infrastructure has raised GDP growth between 2003 and 2023. These effects have been stronger during periods of improved policy coordination and macroeconomic management [15]. While IQ has not been explicitly tested, the post-2003 reforms have highlighted a possible part of governmental improvements in enhancing fiscal policy outcomes.

On the other hand, monetary policy influences economic growth primarily through its channel of policy interest rates [2]. By adjusting these rates, central banks can stimulate investment and consumption or hold inflationary pressures. But then again, its effectiveness critically relies on institutional credibility because central bank independence is important for anchoring inflation expectations to stabilize output [27]. Without strong institutions, monetary policy fails to deliver expected results. Evidently, Havi and Enu [28] have validated that MP operation under weak governing institutions can lead to inflation without real economic gains. Rashid and Husain [29] stated the importance of good institutions to support effectiveness of MP in open and trade-dependent economies. For the case of Thailand, Pastpipatkul and Ko [15] also found that MP has been more effective than FP in sustaining growth, especially during high-growth periods.

Given these points, the interaction between IQ and macroeconomic policy is fundamental to sustaining long-term economic stability and development. Institutions shape the design and outcomes of fiscal and monetary policies, while effective policy implementation, in turn, reinforces institutional strength and forms a condition that supports inclusive, resilient, and sustainable economic growth. But there has still been a clear gap in the existing literature. Few studies have focused on how IQ might affect the effects of fiscal and monetary policy on growth for different samples. Most of them have assumed institutions as fixed or secondary. This study takes a different approach. It uses quarterly time series dataset and advanced econometric methods to examine how IQ shapes policy outcomes in Thailand from Q1: 2003 to Q4: 2023. The findings offer new insights into how institutions shape the link between macroeconomic policy and economic growth for the case of a small open economy like Thailand.

3. Methods of Study

In this study, we have examined the role of IQ strengthening the effect of fiscal-monetary policy on the growth of Thai economy. For this primary objective, we have used the Time Varying Seemingly Unrelated Regression Equations (tvSURE) Model. Regarding financial assets, refs. [30,31] used time-varying VAR models to study time-varying relationships and shown the value of time-varying approaches. Building on this idea, we apply a tvSURE model to examine dynamic policy effects and the role of IQ. Before this, we had verified the reliability of variable inclusion for this analysis by employing a combination of Bayesian Additive Regression Trees (BART) and Bayesian Variable Selection with Shrinking and Diffusing Priors (BASAD) approach. We have employed BART and BASAD to detect the relative importance of institutional and macroeconomic covariates. The important variables have been selected based on the posterior inclusion probability exceeding 0.5 percentile to be consistent with [32]. This methodological design has been chosen to not only address nonlinearity, model uncertainty but also temporal heterogeneity which are prominent in the analysis of policy efficacy.

3.1. Bayesian Additive Regression Trees (BART)

BART, a nonparametric Bayesian ensemble learning model, has approximated an unknown function by summing a set of regression trees. This model was originally proposed by [33], and it has been very useful particularly for nonlinear modeling, data-driven variable selection, and prediction under uncertainty. This model has been applied using the “bartMachine” R package developed by [34] implemented in R version 4.5.1. Assume a regression structure for continuous time series as:

$${{\rm Y}}_i={\mu }_0+{f(x}_i)+{ϵ}_i, {ϵ}_i~N(0,{\sigma }^2)$$

(1)

where the unknown function f(.) is expressed as a sum of H trees with ${\tau }_h$ denoting the binary tree structure and $m_h$ the corresponding leaf parameters:

$$f\left(x\right)={\sum }_{h=1}^{H}g\left(x;{\tau }_h{m}_h\right)$$

(2)

Unlike classical parametric regression models, BART has flexibly learned from the data without imposing a rigid functional form, and regularization has been achieved through the prior distributions on tree depth and leaf parameters which prevents overfitting. Posterior inference has been conducted based on a Metropolis within Gibbs algorithm, and variables with higher posterior inclusion probabilities have been retained from downstream modeling. In this way, BART has ensured that only the most influential covariates are selected in the subsequent structural analysis.

3.2. Bayesian Variable Selection with Shrinking and Diffusing Priors (BASAD)

Following the BART estimation, we apply BASAD model developed by [35] to determine the important input variables within a Bayesian framework. This model is well-suited for high-dimensional regression issues where the number of covariates is large relative to the sample size. The regression model (Equation (3)) is specified with latent binary vector $Z=(Z_1,\dots ,Z_p)\in \{0,1{\}}^p$ such that:

$$\gamma =X\beta +ϵ, ϵ~N\left(0,{\sigma }^2{I}_n\right)$$

(3)

$${\beta }_j|Z_j=\left\{\begin{array}{c}0~N\left(0,{\sigma }^2{\tau }_{0,n}^2\right), if Z_j=0 \left(spike\right)\\ 1~N\left(0,{\sigma }^2{\tau }_{1,n}^2\right), if{ Z}_j=1 \left(slab\right)\end{array}\right.$$

(4)

where $\beta$ is the coefficient vector. And, the prior on the inclusion vector is:

$$Z_j~Bernoulli\left(q_j\right), q_j={\displaystyle \frac{1}{p}}$$

(5)

Shrinking and diffusing priors’ variances depend on the sample size n:

$${\tau }_{0,n}^2=0\left({\displaystyle \frac{1}{n}}\right),{\tau }_{1,n}^2=\mathrm{\Omega }\left({\displaystyle \frac{\mathit{max}\left(p^{2+2\delta },n\right)}{n}}\right)$$

(6)

3.3. Time Varying Seemingly Unrelated Regression Equations (tvSURE)

As we have previously proposed, this study has used a tvSURE model to investigate the role of IQ influencing the dynamic effect of fiscal and monetary policies on economic growth in Thailand. This model is an extension of [36]’s classical SURE model by allowing regression parameters to vary smoothly over time. The tvSURE has been applied using the “tvReg” R package developed by [37]. This model has been a great fit for analyzing the dynamic relationships in macroeconomics and financial systems where the effects of explanatory variables such as fiscal, monetary and institutional proxies are expected to vary dynamically due to policy shifts, structural changes or economic crises. A tvSURE model has consisted of a system of M regression equations each with T observations.

$${\gamma }_{m,t}=x_{m,t}{\beta }_{m,t}+{\epsilon }_{m,t} for m=1,\dots ,M and t=1,\dots ,T$$

(7)

where ${\gamma }_{m,t}$ denotes the dependent variable for equation m at time t, $x_{m,t}$ denotes a vector of explanatory variables for equation m, ${\beta }_{m,t}$ denotes time-varying coefficients and ${\epsilon }_{m,t}$ denotes a zero-mean error term satisfying $E\left[{\epsilon }_{m,t}|x_{m,t}\right]=0$.

The error terms ${\epsilon }_{m,t}$ are allowed to be contemporaneously correlated across equations but are uncorrelated across time. In a matrix form, the full system becomes:

$$Y_{it}=X_{it}\beta \left({\displaystyle \frac{t}{T}}\right)+{\epsilon }_{it}$$

(8)

where $Y_{it}={(y}_{1t},\dots ,x_{Mt}{)}^T$ is the stacked M × 1 vector of dependent variables at time t, $X_t={(x}_{1t},\dots ,x_{Mt}{)}^T$ is a block-diagonal matrix of regressors for each of the M equations, ${\beta }_{it}=({\beta }_{1t}^T,\dots ,{\beta }_{Mt}^T{)}^T$ is the full vector of time varying coefficients function, and the error vector ${\epsilon }_{it}=({\epsilon }_{1t,}\dots ,{\epsilon }_{Mt}{)}^T$ has zero mean and time-varying covariance matrix ${\mathrm{\Sigma }}_t$.

We use a kernel-weighted local least squares specifically time-varying feasible generalized least squares (tvFGLS) estimator to estimate the coefficient function $\beta \left({\displaystyle \frac{t}{T}}\right)$ of the model which minimizes a smoothed sum of squared residuals across time.

$${\widehat{\beta }}_t={\displaystyle \genfrac{}{}{0pt}{}{min}{{\beta }_t}}\sum _{s=1}^T{K}_h({\displaystyle \frac{t-s}{T}})(Y_s-X_s{\beta }_t{)}^T{\mathrm{\Sigma }}_s^{-1}\left(Y_s-X_s{\beta }_t\right)$$

(9)

where $K_h(.)$ is a kernel function (e.g., Gaussian) and h is the bandwidth controlling the degree of smoothing. This method provides consistent and asymptotically normal estimates under standard regularity conditions.

In our empirical analysis, we have estimated a system of two equations: MPWIQ equation for the monetary policy and FPWIQ equation for the fiscal policy. Both equations have included interaction terms between IQ and real GDP growth. This has helped us see how IQ changes the effects of monetary and fiscal policies on growth in Thailand from Q1: 2003 to Q4: 2023. Including these terms has let us capture how institutions and policies work together. The tvSURE model has estimated both equations at the same time. This has improved efficiency and accounted for links between monetary and fiscal policies.

4. Data and Findings

4.1. Data

In this study, we use a quarterly dataset from Q1: 2003 to Q4: 2023 to assess the role of IQ influencing the effect of macroeconomic policies on economic growth in Thailand. We define real GDP growth rate as the dependent variable which is measured as the quarterly percentage change in real gross domestic product which reflects the inflation-adjusted quarterly shift in total economic output. We classify the independent variables into three main categories as IQ indicators, macroeconomic policy tools and control macroeconomic factors. We include six proxies namely government effectiveness, rule of law, political stability, control of corruption, voice and accountability, and freedom of expression to account for IQ. These proxies reflect the strength and inclusiveness of institutions and public trust, all of which have been linked to sustainable growth and policy effectiveness [5,8,38]. We consider policy interest rates variable representing the primary monetary policy instrument and government expenditure serving as the main fiscal policy tool both of which influence aggregate demand and economic stability [39,40]. We also incorporate a wide set of macroeconomic control variables to account for the diverse drivers of Thai economy. Financial conditions are proxied by broad money (M2), exchange rates (USD/THB), consumer price index, unemployment rates which echo monetary transmission, inflation, currency stability and labor market performance [23,41,42,43,44,45]. Additionally, we include household debt along with electricity generation to serve as financial vulnerability and infrastructure quality respectively [46,47]. We further control external sector dynamics and global volatility through foreign direct investment inflows, exports, gold prices, fuel (diesel and gasoline) prices, and a dummy variable accounting for COVID-19 pandemic and 2008 financial crisis [48,49,50]. In order to provide robust findings, we employed a mix of two Bayesian variable selection methods which of both identify the most variables with posterior probabilities exceeding 0.5 on the growth of Thai economy.

4.2. Findings

Bayesian variable selection methods were first applied to find which input variables were most important in explaining economic growth for the case of Thailand from 2003: Q1 to 2023: Q4. We determined which IQ indicators and macroeconomic control variables should be included for the econometric model computation based on the estimated results. The selected variables from both institutions and macroeconomic dimensions were incorporated into the tvSURE model to examine how IQ moderates the relationship between macroeconomic policies and economic growth.

Table 1. Bayesian variable selection results for IQ indicators using BART and BASAD methods.

IQ Determinants	Posterior Probability	BART Selection	IQ Determinants	Posterior Probability	BASAD Selection	Combination *
geffect	0.1775	geffect	vocnacc	0.49	vocnacc	vocnacc
ccorruption	0.177	ccorruption	ccorruption	0.473	ccorruption	ccorruption
vocnacc	0.1628	vocnacc	geffect	0.469
freenexpr	0.1627		freenexpr	0.426
polstable	0.1601		rulenlaw	0.318
rulenlaw	0.16		polstable	0.249

Notes: To ensure robustness, we ran BART model 30 times (replicates) and BASAD model based on 1000 iterations with 500 burn-in and 4 degrees of freedom. Variable with posterior probabilities greater than 0.5 are considered selected at the 95th percentile. * The combination presents a mixed selection method between these two models.

and

Table 2. Bayesian variable selection results for macroeconomic indicators using BART and BASAD methods.

Control Factors	Posterior Probability	BART Selection	Control Factors	Posterior Probability	BASAD Selection	Combination *
logcpi	0.0935	logcpi	logexport	0.641	logexport	logexport
logexport	0.0932	logexport	exts	0.48	exts	loggoldpr
unemp	0.0871	unemp	loggoldpr	0.459	loggoldpr	loghhdebt
loggoldpr	0.0846	loggoldpr	logexr	0.429	logexr	logecg
loggovdebt	0.0821	loggovdebt	loghhdebt	0.424	loghhdebt
loghhdebt	0.0789	loghhdebt	logecg	0.397	logecg
logecg	0.0757	logecg	loggovdebt	0.362	loggovdebt
logmsupply2	0.0709		logmsupply2	0.361
gasolinepr	0.068		logcpi	0.339
dieselpr	0.0657		dieselpr	0.304
logexr	0.0558		loggovrev	0.287
fdi	0.051		gasolinepr	0.239
exts	0.0505		unemp	0.23
loggovrev	0.043		fdi	0.138

Notes: To ensure robustness, we ran BART model 30 times (replicates) and BASAD model based on 1000 iterations with 500 burn-in and 4 degrees of freedom. Variable with posterior probabilities greater than 0.5 are considered selected at the 95th percentile. * The combination presents a mixed selection method between these two models.

list the selected variables identified by the BART model, BASAD model, and a combination approach.

4.2.1. Variable Selection Results by Using BART and BASAD Methods

This study identified voice and accountability and control of corruption as the most relevant and important IQ indicators (see Table 1, combination results). The importance of voice and accountability and control of corruption reveals the prominent role of governance in sustaining growth. These two IQ proxies directly influence the integrity of policies and regulations as well as sustainable growth for the period between 2003: Q1 and 2023: Q4. This finding discloses that improvements in securing political inclusiveness and lowering corruption are essential to promote the efficacy of macroeconomic policies for the Thai economy. On the macroeconomic side, the variables log exports, log gold prices, log household debt and log electricity generation emerged as key control factors for this study (see Table 2, combination results). The combination selection findings of the importance of exports reaffirm the status of Thailand as an open economy reliant on global trade. The finding of importance of gold prices on growth indicates the investor sentiment to commodity price fluctuations. The finding of importance of household debt emphasizes the dual role of credit conditions because excessive debt may weigh on financial instability while moderate household indebtedness fuels consumption resulting in growth in the short term. Lastly, the importance of electricity generation underscores the need for reliable energy supply to not only support the quality of daily life but also industrial and total productivity. In sum, these estimated results showed that IQ and macroeconomic fundamentals jointly associated with the performance of the Thai economy. The final selected IQ proxies were multiplied by real GDP growth for the robustness of their moderating effects on the policy-growth nexus.

4.2.2. tvSURE Estimation Results Without and with Institutional Quality Proxies

This section discusses the empirical findings of the study on how IQ influenced macroeconomic policy-driven growth in Thailand over the period from Q1: 2003 to Q4: 2023, based on tvSURE model estimations. The results were visualized in Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15 and Figure 16, and all figures incorporated 90 percent bootstrap confidence intervals based on 100 runs to ensure robust findings. The analysis was disaggregated into monetary policy (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8), denoted MPWIQ, and fiscal policy (Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15 and Figure 16), denoted FPWIQ. The final selected IQ proxies by the combination of BART and BASAD models were transformed into interaction terms (IQ proxies × real GDP growth). The inclusion of interaction terms explores their moderating effects on the growth–macroeconomic policy correction.

Policy interest rates (interest) positively affected economic growth during around Q2: 2013 to Q4: 2013 and again around Q4: 2014 to Q1: 2015 with effects between 0.1 and 0.2 (Figure 2). Outside these periods, their impact was mostly neutral. The interaction term of voice and accountability (fvocnacc) showed large fluctuations over the years, with some quarters sustaining growth and other times slowing it (Figure 3). The second interaction term of IQ proxy, control of corruption, generally had a negative effect on growth throughout the period except for a brief positive influence around Q1:2017 (Figure 4). Exports mostly supported growth except during around Q2: 2014 to Q4: 2014, Q4:2015 and late 2021 to 2023 when their effects weakened (Figure 5). Gold prices negatively affected growth in 2007, 2009, and again in late 2022 (Figure 6). Household debt increased its positive impact on growth until about Q1: 2019 then stabilized, while electricity generation dipped mid-period but rose again by Q4: 2023. This indicates increasing support for growth. On the other hand, fiscal policy (loggovexp) showed a different pattern. It remained frequently neutral in its effect on growth throughout the entire period from Q1: 2003 to Q4: 2023 with minimal fluctuations. The interaction term of IQ proxy (voice and accountability) displayed strong fluctuations with positive effects around Q4: 2007 to Q2: 2008, Q2: 2012 to Q3: 2012, and Q2: 2022 to Q2: 2023, while showing negative effects during other times. The other interaction term of IQ proxy, control of corruption, consistently showed a negative relationship with growth except for a brief positive effect around Q1: 2017. Exports (Figure 13) mostly supported growth but weakened during Q1: 2014 to Q4: 2014, Q1: 2017 to Q2: 2017 and Q2: 2022 to Q4: 2023. Gold prices (Figure 14) also had a slight negative impact on growth with a more noticeable drop to about −0.3 in the last quarters of the study around Q2: 2022 to Q4: 2023. In sum, these results showed that monetary policy worked differently over time. Interest rates helped growth only in some periods. Policy effects changed with Thailand’s political and institutional shifts. The strong positive impact in Q2: 2013–Q1: 2015 came after political stabilization following the 2011 floods. It occurred before the 2014 military coup. Sharp changes in IQ proxies during 2006 and 2014 match the coups in those years. These events disrupted governance continuity. Voice and accountability improved briefly in 2012 and 2022–2023. These gains linked to constitutional reforms and decentralization initiatives. Control of corruption turned positive around 2017. This matched major anti-corruption campaigns. IQ fell during unrest such as the 2010 protests. This weakened policy transmission. External shocks and shifts in local confidence also shaped results. Institutional effects were not stable over time.

We also found that individual coefficient paths vary across Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11, Figure 12, Figure 13, Figure 14, Figure 15, Figure 16, Figure 17, Figure 18, Figure 19, Figure 20, Figure 21, Figure 22, Figure 23, Figure 24, Figure 25, Figure 26, Figure 27 and Figure 28. Several broader patterns appear. Monetary policy interest rates had positive effects only in short periods. For most of the time, the effect was neutral. This means their growth impact in Thailand was episodic, not constant. Fiscal policy effects stayed close to zero, with or without IQ proxies. This shows a generally neutral role. Institutional quality interactions were unstable. Voice and accountability switched between positive and negative effects. Control of corruption was mostly negative in the short run. This may reflect the costs of reform. The same macroeconomic controls like exports, gold prices, household debt, and electricity generation which were important in both policy areas. Their influence also changed with global and domestic shocks. The growth, policy and institution links are highly dynamic. Moreover, growth in exports improves growth but the effect is time varying. This shows risk from shocks of global trade. Gold prices fall when growth rises. During crises, this may show fear, while rising household debt tends to have a positive effect on growth. Electricity generation may reveal the quality of infrastructure. Its rise comes late but may promote long-term growth. Fiscal policy shows weak time effects. This may mean poor systems or slow results. Its link with institutions also changes. This may indicate a need for better governance institutions. Overall, stronger policy coordination is needed. Monetary, fiscal, and institutional tools must work together to support growth. These quarterly findings from Q1:2003 to Q4:2023 provide a broad understanding of the role of governance in economic resilience and effective policy design in Thailand for long-run and sustainable economic growth. This could be similar in emerging economies, as Thailand is an emerging economy.

In addition, the comparison between models with and without the moderating role of IQ proxies showed some changes in the estimated effects of macroeconomic policies and economic growth. The tvSURE model estimates without IQ proxies revealed that policy interest rates (Figure 18) had significant negative effects on growth around Q2: 2013 to Q1: 2015. The effect fluctuated throughout the study period. In the model with IQ proxies (Figure 2), this effect was flatter and remained close to zero throughout the sample quarters. In fact, it positively affected economic growth especially during around Q2: 2013 to Q4: 2013 and again around Q4: 2014 to Q1: 2015 with effects between 0.1 and 0.2 (Figure 2). Similarly, fiscal policy measured by government expenditure (Figure 24) in the model without IQ proxies had large and consistent negative effects especially from Q2: 2010 to Q4: 2018. When IQ proxies were included (Figure 10), the effects of government expenditure exposed a different pattern. It remained frequently neutral, exhibiting around zero in its effect on economic growth from Q1: 2003 to Q4: 2023 with very low fluctuations. Overall, models excluding IQ overstate the size and stability of policy effects. Including IQ proxies yields more realistic and nuanced estimates. This has validated that strong institutions are crucial for sustainable and effective macroeconomic policy.

Interestingly, the control of corruption shows a negative or unstable effect on growth in the short run. This may be due to the costs of reform. Fighting corruption can disrupt the economy and face resistance from powerful groups. These changes take time to show positive effects. Thailand’s political history adds to the challenge. The country has faced coups, constitutional changes, and unrest. Political instability makes it hard to enforce reforms. It creates uncertainty for investors and policymakers. This slows growth and weakens institutions Therefore, the negative impact reflects these short-term disruptions. It also shows how political instability affects governance. Strong institutions help growth, but only with stable politics and support for reform.

These findings support the importance of IQ in shaping the effectiveness of macroeconomic policies. Moreover, models that omit IQ proxies tend to exaggerate both the magnitude and stability of policy effects. These findings confirm that the inclusion of institutions leads to more realistic estimates. This study thus concludes that strong institutions are essential for effective public policy performance and sustainable growth for the Thai economy between 2003 and 2023. Besides, monetary and fiscal policies alone are insufficient without supportive governance. Addressing political instability and improving reform implementation are key to unlocking sustainable growth.

To check the robustness, we employed a vector autoregression with exogeneous variables (VARX) model. Sims [51] first proposed the model VAR and it has been widely used to study dynamic associations between macroeconomic variables and policy shocks. The VAR model allows adding exogenous variables to the system considering the outside influences. Important studies using VARX include [52,53]. We choose this model because it is effective to examine the interactions of fiscal and monetary policies, IQ, and growth well. It serves as a good check alongside the main tvSURE model. For this, we used the same variables selected by the coordination of BART and BASAD approach. The VARX results support the main findings. Macroeconomic policies and institutions jointly affect economic growth in Thailand. Interest rates respond to changes in both GDP growth and IQ. This confirms the interaction effects found in the main model. Government spending shows inertia. This supports the neutrality of fiscal policy in the tvSURE estimates. Exogenous variables such as exports and household debt also play a role. They influence policy responses, especially across different periods. This model performs well and reveals consistent patterns (Appendix C). Impulse response functions (IRFs) are reported in Appendix D. These IRFs present how policies react to shocks in GDP growth and vice versa. Government spending declines in the first two quarters. It turns positive after the fourth quarter. Policy interest rates fall slightly at first. But the effect remains near zero afterward. These results suggest moderate policy responses to output shocks. There is little evidence of strong fiscal or monetary expansions (Appendix D).

5. Conclusions, Implications and Limitations

This study investigates the time-varying effects of fiscal and monetary policy on economic growth in Thailand for the period from Q1:2003 to Q4:2023. A system of two equations is estimated using the tvSURE model for the moderating role of IQ on the effectiveness of policy implementation on sustainable growth. This study shows that IQ proxies like voice and accountability, and control of corruption are key determinants of how well fiscal and monetary policies support economic growth in Thailand from 2003 to 2023. Using the Time-Varying Seemingly Unrelated Regression Equations (tvSURE) model with Bayesian variable selection methods (BART and BASAD), we find these institutional factors are the most important variables in strengthening and stabilizing the effect of monetary and fiscal policies on growth. Macroeconomic controls such as exports, gold prices, household debt, and electricity generation also play significant roles.

The main results show that policy impacts vary over time. Interest rates positively influence growth during specific periods (notably around 2013 and 2014–2015) but remain neutral at other times. Voice and accountability demonstrate fluctuating effects on growth, and control of corruption mostly exerts a negative influence except for a brief positive phase around early 2017. Exports consistently support growth though their impact weakens during certain quarters. Gold prices have a negative effect during global market shocks. Household debt boosts growth until around 2019 before stabilizing, while the positive effect of electricity supply rises at the end of the study period. Fiscal policy, measured by government expenditure, shows minimal direct impact on growth with effects remaining largely neutral throughout the study period. The negative and variable effects of corruption control highlight potential governance challenges. These findings emphasize the essential role of strong institutions in enhancing the efficacy of macroeconomic policies. Greater voice and accountability and strong control of corruption can stabilize policy impacts and support economic resilience. The comparison also reveals that excluding IQ proxies results in overstated and more volatile policy impacts. This study adds empirical evidence about strong IQ on the economic resilience in emerging markets.

The findings of this study have practical implications for the Thai government and policymakers. They highlight the necessity to adopt appropriate tools and frameworks to strengthen institutions. Improving political inclusiveness and enhancing efforts to control corruption can maximize the effectiveness of macroeconomic policies. Stabilizing the positive effects of interest rates and export growth depends on a strong governance framework. The negative trends related to household debt and electricity generation signal the importance of careful debt management and investment in energy infrastructure supported by robust institutions. Even though these findings are specific to Thailand, they may also be relevant to other emerging economies with similar encounters. Besides, our results are consistent with studies in other emerging markets including [8,15,46,47]. Our study adds further evidence that strong institutions influence how fiscal and monetary policies affect real growth from time to time. Moreover, the changes in institutional quality effects match [5]’s theory of shifts from extractive to inclusive institutions. Periods of stronger governance, such as reform phases, show partial moves toward inclusiveness. These shifts help policies work better and more efficiently. Conversely, times of political instability and weak governance reflect extractive institutions. Here, elite control and less accountability reduce policy effectiveness. These reveal that institutions in Thailand are neither fixed nor linear.

In this study, the analysis only includes the selected variables from a mix-order approach of BART and BASAD for accounting IQ and macroeconomic fundamentals. This may not perfectly echo the theoretical constructs. Future research could benefit from considering other factors. The tvSURE model is suitable for analyzing a system of equations with time-varying relationships across multiple policy instruments, offering flexibility in capturing dynamic interactions without imposing strong structural assumptions. So, using econometric alternatives to the tvSURE model in exploring the nonlinear relationships may provide further insights into the policy-growth nexus, as the performance of this model may depend on the chosen variables and its specification. This study does not confirm causal direction but time-varying associations between fiscal and monetary policies, IQ and economic growth. Therefore, the conclusions and implications should be interpreted with caution. Even though the use of quarterly data studies the short-run dynamics effectively, future research may benefit from particularly applying models which are capable of capturing both short-/long-run relationships such as nonlinear ARDL. These findings are limited to the case of the Thai economy for the specific periods. The complex effects involving electricity generation suggest that policies must fit each country’s governance strengths and weaknesses. Future studies should test this model in other Southeast ASEAN countries by adding more institutional factors and look at possible nonlinear effects using advanced methods. The negative trend in electricity generation’s impact after 2017 needs further study to conduct, including energy policies and green technologies. Also, reverse causal relationships may exist between economic growth IQ. Our model does not address this. Future research should explore this issue.