Impact of Fiscal and Monetary Policy on Unemployment in Malaysia: ARDL Approach

Impact of Fiscal and Monetary Policy on Unemployment in Malaysia: ARDL Approach

RHM. Radzil^1*, IM. Hashim¹, R. Razlina²

^*1Faculty of Business and Management, Universiti Teknologi MARA, 27600 Raub, Pahang

²Faculty of Management and Informatic, Universiti Islam Pahang Sultan Ahmad Shah, 25150 Kuantan, Pahang

*Corresponding Author

DOI: https://dx.doi.org/10.47772/IJRISS.2025.909000655

Received: 08 October 2025; Accepted: 16 October 2025; Published: 25 October 2025

ABSTRACT

This study investigates the impact of proxy for fiscal and monetary policies on unemployment in Malaysia, utilizing annual time series data from 1995 to 2021. Employing an econometric approach, the research primarily focuses on diagnostic tests of a hypothesized unemployment model, including unit root tests, ARDL bounds testing for cointegration, and tests for heteroskedasticity and serial correlation. The Phillips-Perron (PP) unit root tests reveal that most variables exhibit stationarity at first difference, confirming their integration order. The ARDL Bounds Test indicates the presence of a long-run relationship among the variables, suggesting that fiscal and monetary policy instruments, alongside unemployment, are cointegrated over the long term. A key component of this analysis involves a regression of the squared residuals (RESID2) from a primary unemployment model, which serves as a diagnostic test for heteroskedasticity. The results from this regression show that lagged unemployment rates and lagged interest rates, as well as current government tax revenue and grants, significantly influence the variance of the unemployment model’s residuals. Specifically, UNEMPRATE(-2) and IR(-2) exhibit significant coefficients, indicating that past unemployment and interest rate volatility contribute to the current volatility of the model’s errors. The Breusch-Godfrey Serial Correlation LM Test further confirms the absence of serial correlation in the model. These findings underscore the complex interplay between macroeconomic policies and labor market stability in Malaysia, highlighting the importance of considering residual variance in policy formulation. The study concludes with policy implications for enhancing the effectiveness of fiscal and monetary measures in mitigating unemployment in Malaysia.

Keywords: Fiscal policy, Monetary policy, Unemployment, Malaysia, ARDL, Heteroskedasticity, Serial Correlation, Time series analysis.

INTRODUCTION

Unemployment stands as a formidable and persistent challenge for economies worldwide, transcending geographical and developmental boundaries. Its multifaceted nature, encompassing economic, social, and political dimensions, renders it a central concern for policymakers, particularly during periods of economic instability or crisis. A nation’s ability to maintain a high level of employment is intrinsically linked to its economic growth trajectory, as underutilized labor resources represent a significant drag on potential output and overall prosperity (Onwuka, 2022). The pursuit of full employment is thus a cornerstone of sound macroeconomic management, signifying the optimal utilization of available human capital.

In addressing the intricate dynamics of unemployment, two primary macroeconomic policy tools emerge as pivotal: fiscal policy and monetary policy. Fiscal policy, broadly defined as the government’s strategic deployment of taxation and public expenditure, exerts its influence by directly modulating aggregate demand within an economy. Expansionary fiscal measures, characterized by increased government spending or reductions in tax burdens, are designed to stimulate economic activity, thereby fostering job creation and alleviating unemployment. Conversely, contractionary fiscal policies, often implemented to curb inflation or reduce national debt, can lead to a dampening of economic activity and, consequently, an increase in unemployment (Onwuka, 2022). The effectiveness of fiscal policy, however, is not uniform, varying significantly with the specific instruments employed and the prevailing economic context. For instance, capital expenditures, such as investments in infrastructure, have been shown to generate more sustainable employment than recurrent spending or broad tax cuts (Mahmood & Sadiq, 2020).

Monetary policy, on the other hand, falls under the purview of central banks and involves deliberate actions to manage the quantity and availability of money in circulation. This is primarily achieved through adjustments to interest rates, reserve requirements, and open market operations. The overarching aim of monetary policy is to influence economic activity, often with dual objectives of price stability and full employment (Onwuka, 2022). By lowering interest rates, for example, central banks can reduce borrowing costs for businesses and consumers, stimulating investment and consumption, which in turn can lead to increased production and employment. However, the impact of monetary policy on unemployment can be complex, with potential for lags and varying degrees of effectiveness depending on the responsiveness of economic agents and the presence of other macroeconomic distortions.

Malaysia, as a dynamic economy within the ASEAN region, has consistently grappled with the challenge of unemployment, particularly in the face of global economic shocks and periods of structural adjustment. The efficacy of fiscal and monetary interventions in the Malaysian context is a subject of ongoing debate and empirical scrutiny. While public spending on infrastructure and targeted employment programs has historically been associated with short-term reductions in unemployment (Saari et al., 2018; Abd Majid & Rasiah, 2014), the long-term sustainability and broader impact of these measures warrant closer examination. Similarly, the role of Bank Negara Malaysia’s monetary policy in fostering labor market stability, through mechanisms such as interest rate adjustments and credit availability, remains a critical area of inquiry (Pham, 2022).

This study aims to contribute to the existing body of knowledge by empirically examining the impact of fiscal and monetary policy instruments on unemployment in Malaysia. Drawing upon a comprehensive dataset and employing rigorous econometric techniques, the research seeks to uncover the intricate relationships between key macroeconomic variables and labor market outcomes. While the primary focus of the provided results pertains to diagnostic tests of a hypothesized unemployment model, including tests for heteroskedasticity and serial correlation, these diagnostics offer crucial insights into the stability and reliability of the underlying relationships. Understanding the volatility of unemployment model residuals, as explored through the regression of squared residuals, is paramount for robust policy formulation, as it sheds light on the predictability and consistency of policy effects.

LITERATURE REVIEW

The interplay between fiscal and monetary policies and their impact on unemployment has been a cornerstone of macroeconomic theory and empirical research for decades. This section delves into the theoretical underpinnings and reviews key empirical findings, with a particular focus on the Malaysian and broader ASEAN contexts.

Fiscal Policy and Unemployment

Fiscal policy, through its direct influence on aggregate demand, has long been recognized as a potent tool for addressing unemployment. Expansionary fiscal measures, such as increased government spending on infrastructure projects or social welfare programs, directly create jobs and stimulate economic activity. Tax cuts, by increasing disposable income, can boost consumption and investment, indirectly leading to job creation.

Empirical evidence from developed economies consistently supports the notion that fiscal consolidations tend to exacerbate unemployment. Turrini (2013), in a study of European Union countries, found that austerity measures were associated with higher unemployment, particularly in labor markets characterized by rigidity. His panel regression analysis highlighted those countries with stringent employment protection legislation experienced more prolonged unemployment spells following fiscal austerity. Similarly, Monastiriotis and Antoniou (2013) utilized a Structural Vector Autoregression (SVAR) model to analyze fiscal shocks in Greece, concluding that a 1% reduction in government spending could lead to a 0.91 percentage point increase in unemployment over a two-year period. These findings underscore the critical importance of countercyclical fiscal interventions during economic downturns to mitigate job losses. Further, Gechert and Rannenberg (2015), employing a panel VAR approach for OECD countries, demonstrated that government expenditure generally yields stronger employment multipliers compared to tax reductions, emphasizing the efficacy of targeted spending in reducing unemployment.

In the context of developing countries, research often echoes these findings but also highlights the nuances of fiscal policy effectiveness. Mahmood and Sadiq (2020), applying an Autoregressive Distributed Lag (ARDL) model to Nigerian data, found that capital expenditure significantly reduced unemployment, whereas recurrent spending and higher taxes had an exacerbating effect. This suggests that the composition of fiscal spending matters, with productive investments offering more sustainable job creation.

Within the Malaysian context, fiscal policy has played a prominent role in addressing unemployment, particularly during periods of global financial distress. Public spending, especially on infrastructure development and employment-generating programs, has been consistently linked to short-term reductions in unemployment. Saari et al. (2018), using a Computable General Equilibrium (CGE) model, found that increased government spending on public infrastructure in Malaysia boosted job creation, particularly within the services and construction sectors. Likewise, Abd Majid and Rasiah (2014) argued that targeted fiscal measures, such as wage subsidies and public sector hiring initiatives, proved effective in mitigating unemployment during the 2008 global financial crisis. These studies highlight the Malaysian government’s proactive approach in deploying fiscal tools to stabilize the labor market.

Across the broader ASEAN region, fiscal multipliers tend to be moderate to high, indicating that government expenditure has a substantial impact on aggregate demand and, consequently, on employment (Ginting & Aji, 2020). However, the effectiveness of these measures is often contingent on the specific economic conditions, the quality of governance, and the efficiency of public spending.

Monetary Policy and Unemployment

Monetary policy, primarily managed by central banks, influences unemployment through its effects on interest rates, credit availability, and overall economic activity. Lower interest rates reduce the cost of borrowing for businesses, encouraging investment in new projects and expansion, which in turn leads to increased demand for labor. Similarly, lower interest rates can stimulate consumer spending, further boosting aggregate demand and employment.

Several studies have explored the impact of monetary policy on unemployment. Stockhammer and Sturn (2011) investigated the impact of monetary policy on unemployment hysteresis, finding that accommodative monetary policies can help prevent persistent unemployment. The International Monetary Fund (IMF) (2022), in its policy reports, has often highlighted how liquidity support measures, especially during crises, can stabilize Small and Medium-sized Enterprise (SME) employment, underscoring the complementary role of monetary tools to fiscal initiatives. Fukao (2020) observed that monetary easing in Japan increased labor flexibility, though its impact on productivity remained debated, suggesting that monetary policy needs to be complemented by long-term structural reforms for full effectiveness. In China, He and Liu (2022) found that targeted monetary easing significantly supported employment in manufacturing and service sectors post-COVID-19, advocating for a focus on labor-intensive sectors.

In Malaysia, Bank Negara Malaysia (BNM) plays a crucial role in managing monetary policy to achieve price stability and support sustainable economic growth, which indirectly impacts employment. BNM’s annual reports often detail its strategies for maintaining financial stability and fostering an environment conducive to job creation, through measures such as adjusting the Overnight Policy Rate (OPR) and managing liquidity in the financial system. Changes in the OPR influence commercial bank lending rates, affecting investment and consumption decisions, and ultimately, employment levels.

Pham (2022), in a study focusing on ASEAN economies using a panel VAR approach, found evidence that monetary policy significantly influences unemployment rates in the region. The study suggests that interest rate adjustments and other monetary instruments can effectively mitigate unemployment, particularly during periods of financial disruption. However, the effectiveness of monetary policy in addressing structural unemployment issues, such as skill mismatches or labor market rigidities, is often limited, requiring a coordinated approach with fiscal and structural policies.

Policy Coordination and Challenges

The literature consistently emphasizes that the optimal approach to tackling unemployment often involves a coordinated effort between fiscal and monetary policies. Kuttner and Posen (2019), in their evaluation of Abenomics in Japan, highlighted that coordinated fiscal and monetary policies significantly improved employment outcomes, particularly for women and the elderly, reinforcing the idea that joint policy initiatives enhance labor force participation.

Despite the potential for synergy, several challenges can impede the effectiveness of these policies. These include macroeconomic instability, political constraints, implementation lags, and structural issues within labor markets. For instance, skill mismatches between the available workforce and industry demands, as well as the prevalence of informal employment sectors, can limit the impact of even well-intentioned fiscal and monetary interventions (Arifin & Ismail, 2023).

Both fiscal and monetary policies have demonstrated varying degrees of success in combating unemployment in Malaysia and the broader ASEAN region. While fiscal measures often offer more immediate and direct employment benefits, monetary policy complements these efforts, especially during periods of financial instability. The ultimate effectiveness of these policies is contingent upon a stable macroeconomic environment, robust policy coordination, and the capacity to address underlying structural labor market rigidities. The empirical analysis in the subsequent sections will provide specific insights into these relationships within the Malaysian context.

METHODOLOGY

This study employs a quantitative research design utilizing time series econometric methods to analyze the impact of fiscal and monetary policy variables on unemployment in Malaysia by Autoregressive Distributed Lag (ARDL). The analysis spans the period from 1995 to 2021, encompassing 27 annual observations. Pre-testing such as stationary test conducted using Philip-Perron (PP). The test is to ensure variables are either integrated of order zero (I(0)) or one (I(1)). This is because the ARDL model is flexible and can handle a mix of stationary and non-stationary (but not second-order integrated, I(2) (Pesaran and Shin,1999).

Data Sources and Variables

The data for this study are annual time series secondary data retrieved from World Bank Data. The variables considered in the analysis, as indicated are defined as follows:

• Unemployment, total (% of total labor force) (national estimate). This indicator measures the share of the labor force that is without work but available for and actively seeking employment (World Bank, 2024). It captures the efficiency of the labor market and serves as the primary dependent variable in this study. While unemployment itself is not a policy instrument, it is a key target variable for both fiscal and monetary policies. Expansionary monetary and fiscal policies that stimulate aggregate demand typically reduce unemployment (Blanchard, 2017).
• Interest rate spread (lending rate minus deposit rate, %). The interest rate spread represents the difference between the rates charged by banks on loans and those paid on deposits, reflecting the efficiency of financial intermediation (World Bank, 2024). It is classified under monetary policy. A wider spread implies higher intermediation costs or financial sector inefficiencies, which may discourage borrowing and investment, thereby increasing unemployment. Conversely, a narrower spread enhances credit access and investment, potentially reducing unemployment (Mishkin, 2019).
• Government revenues of tax revenue. Tax revenue refers to the funds collected by the government through various forms of taxation, including income, corporate, and consumption taxes (World Bank, 2024). It is a fiscal policy variable that indicates the government’s ability to finance expenditures. Higher tax revenue can support public spending and reduce unemployment by stimulating demand, though excessive taxation may suppress private investment and increase unemployment (Keynes, 1936; Romer, 2019).
• Revenue – grants. Grants represent non-repayable transfers received from foreign governments or international organizations (World Bank, 2024). They form part of fiscal policy, enhancing the government’s spending capacity without increasing the domestic tax burden. Such external fiscal support can finance employment-generating programs and social spending, thereby reducing unemployment, especially in developing economies (Todaro & Smith, 2020).
• General government final consumption expenditure (constant 2015 US$). This is another fiscal policy instrument. The variable captures government spending on goods and services, including employee compensation and operational costs, but excludes investment expenditures (World Bank, 2024). As a fiscal policy instrument, an increase in government consumption stimulates aggregate demand, output, and employment. Therefore, higher government consumption is generally associated with lower unemployment (Auerbach & Gorodnichenko, 2012).
• GDP (current USD-productivity value). Gross Domestic Product (GDP) measures the total market value of goods and services produced within an economy at current prices (World Bank, 2024). Although GDP is not a direct policy tool, it reflects the macroeconomic outcome of fiscal and monetary interventions. Higher GDP or stronger economic growth increases labor demand and consequently reduces unemployment, in line with Okun’s Law (Okun, 1962).
• Official exchange rate (LCU per US$, period average). This variable represents the average number of local currency units (LCU) required to purchase one U.S. dollar during a specific period (World Bank, 2024). It falls under monetary and external policy. A depreciation of the domestic currency improves export competitiveness and stimulates production, potentially reducing unemployment. Conversely, an appreciation may weaken export sectors and increase unemployment (Krugman & Obstfeld, 2018).
• Broad money to total reserves ratio. This ratio compares the broad money supply (M2) with total foreign reserves (World Bank, 2024). It is a monetary indicator reflecting liquidity and financial depth relative to reserve adequacy. A higher ratio indicates greater liquidity and credit potential, which can support investment and job creation. However, an excessively high ratio may signal financial instability that could eventually raise unemployment (Friedman, 1968).

While the main objective of the paper is to examine the impact of fiscal and monetary policy on unemployment, the provided results focus on the diagnostic properties of model. Therefore, the methodology and results sections will interpret the provided output in this context, inferring the characteristics of the underlying model.

Econometric Model and Techniques

The econometric analysis in this study involves several stages, including unit root testing, cointegration analysis, and diagnostic tests for model validity.

Unit Root Tests

Before proceeding with regression analysis, it is crucial to ascertain the stationarity properties of the time series variables. Non-stationary series can lead to spurious regression results. The Phillips-Perron (PP) unit root test is employed to examine the order of integration of each variable. The PP test is robust to serial correlation and heteroskedasticity in the error term. The test is conducted for each variable at level and at first difference, with and without a constant and trend, to determine their stationarity. The null hypothesis of the PP test is that the series has a unit root (it is non-stationary).

ARDL Bounds Test for Cointegration

Given that the unit root tests may indicate a mix of I(0) and I(1) variables (stationary at level and first difference, respectively), the Autoregressive Distributed Lag (ARDL) bounds testing approach to cointegration, developed by Pesaran, Shin, and Smith (2001), is suitable. This approach allows for the investigation of long-run relationships among variables regardless of whether they are I(0) or I(1), but not I(2) or beyond. The ARDL bounds test involves estimating an unrestricted error correction model (UECM) and testing the joint significance of the lagged levels of the variables in the conditional ARDL error correction model using an F-statistic. The null hypothesis of no long-run relationship is tested against the alternative hypothesis of a long-run relationship. The decision is made by comparing the calculated F-statistic with critical values provided by Pesaran et al. (2001).

Diagnostic Tests: Heteroskedasticity and Serial Correlation

The provided results specifically include diagnostic tests that are critical for assessing the reliability and efficiency of the estimated model.

• Heteroskedasticity Test

The regression is a variant of the Breusch-Pagan or White test for heteroskedasticity. This test examines whether the variance of the residuals from a primary regression model (where unemployment rate is the dependent variable) is constant across all observations. If heteroskedasticity is present, the standard errors of the estimated coefficients in the primary model will be biased, leading to incorrect inferences about their statistical significance. A statistically significant coefficient for any of the independent variables in this regression indicates that the variance of the residuals is related to that variable, suggesting the presence of heteroskedasticity. The F-statistic and Obs*R-squared from this regression are used to test the null hypothesis of homoskedasticity (constant variance).

• Breusch-Godfrey Serial Correlation LM Test

The Breusch-Godfrey Serial Correlation LM Test is used to detect the presence of autocorrelation (serial correlation) in the residuals of the primary model. Autocorrelation violates the assumption of independent error terms, leading to inefficient parameter estimates and biased standard errors.

The test involves regressing the residuals from the primary model on the original independent variables and lagged residuals. The null hypothesis is that there is no serial correlation up to a specified lag order. The F-statistic and Obs*R-squared from this test are used to make inferences. A high p-value (typically greater than 0.05) for these statistics indicates that the null hypothesis of no serial correlation cannot be rejected. By conducting these diagnostic tests, the study ensures that any conclusions drawn about the impact of fiscal and monetary policies on unemployment are based on a statistically robust and well-specified econometric model.

RESULTS

This section presents the empirical findings from the econometric analysis, including the unit root tests, ARDL bounds test for cointegration, and the diagnostic tests for heteroskedasticity and serial correlation. 

Unit Root Test Results (Phillips-Perron)

The stationarity of the time series variables was assessed using the Phillips-Perron (PP) unit root test. The results for each variable at level and at first difference, considering models with a constant, with a constant and trend, and without a constant and trend, are summarized in Table 1.

Table 1: Unit Root Test Table (Phillips-Perron)

*Notes: *, *, *** denote stationarity at 10%, 5%, and 1% significance levels, respectively. “n0” indicates non-stationarity.

The results indicate that at level, only ‘BM’ (Broad Money) is stationary at the 5% significance level when tested with a constant. ‘UNEMPRATE’ shows weak evidence of stationarity at the 10% level with a constant, but is non-stationary under other specifications. All other variables (GDP, IR, BMTR, GE, GRTAX, GRGRANTS, ER) are found to be non-stationary at their levels across all test specifications.

However, upon first differencing, all variables (d(UNEMPRATE), d(GDP), d(IR), d(BM), d(BMTR), d(GE), d(GRTAX), d(GRGRANTS), d(ER)) become stationary at the 1% significance level when tested with a constant. This implies that most variables are integrated of order one, I(1), while BM is I(0). The mixed order of integration (I(0) and I(1)) confirms the appropriateness of using the ARDL bounds testing approach for cointegration analysis.

ARDL Bounds Test for Cointegration

The ARDL Bounds Test was conducted to investigate the presence of a long-run cointegrating relationship among the variables. The F-statistic obtained from the test is compared against critical values to determine cointegration.

Table 2: ARDL Bounds Test Results

Table 2 presents the results of the ARDL bounds test. The computed F-statistic value is 9.304, which is greater than the upper critical bound value (I(1)) at all conventional significance levels ; 10% (3.13), 5% (3.50), 2.5% (3.84), and 1% (4.26). Since the calculated F-statistic exceeds the upper bounds across all levels of significance, the null hypothesis of no long-run relationship among the variables is rejected.

This finding confirms the presence of a long-run cointegrating relationship between unemployment and its selected macroeconomic determinants (interest rate spread, tax revenue, grants, government consumption expenditure, GDP, exchange rate, and broad money to total reserves ratio). The result implies that these variables move together in the long run, and deviations from the long-run equilibrium are temporary and self-correcting over time.

The confirmation of cointegration supports the estimation of both long-run coefficients and the short-run dynamics through the Error Correction Model (ECM). The presence of cointegration suggests that fiscal and monetary policy variables jointly exert significant long-run influences on unemployment, underscoring the interdependence between macroeconomic stability and labor market outcomes in the Malaysian context.

Error Correction Model (ECM) : Short Run and Long Run

Table 3: Short Run and Long Run Relationship

The ARDL estimation results reveal both short-run dynamics and long-run equilibrium relationships between unemployment and selected macroeconomic variables in Malaysia for the period 1995 to 2021. The long-run results show that interest rate spread (IR), government expenditure (GE), and Gross Domestic Product (GDP) significantly influence unemployment. Specifically, the interest rate exhibits a positive and significant coefficient (0.3006, p = 0.0089), suggesting that higher lending rates raise borrowing costs, discourage investment, and consequently increase unemployment (Mishkin, 2019). Government expenditure also shows a positive and significant coefficient (0.9287, p = 0.0036), implying that increased public consumption may not necessarily translate into job creation. This could indicate that government spending in Malaysia is primarily consumption-based rather than investment-driven, potentially crowding out private sector employment (Auerbach & Gorodnichenko, 2012). Conversely, GDP exerts a negative and significant effect (- 0.3671, p = 0.0177), consistent with Okun’s Law, confirming that economic expansion lowers unemployment through increased labor demand.

In the short run, government grants, GDP, and exchange rate are significant determinants of unemployment. Higher grants (- 0.1986, p = 0.0372) reduce unemployment by supporting development projects and social programs. Similarly, GDP growth (- 0.6793, p = 0.0048) and exchange rate depreciation (- 0.9758, p = 0.0054) contribute to short-term job creation by stimulating domestic output and export competitiveness. The error correction term (-1.8505, p = 0.0005) is negative and highly significant, confirming a strong adjustment mechanism that corrects deviations from equilibrium within a year. In short, the results highlight that both monetary and fiscal factors shape unemployment trends in Malaysia. Effective management of interest rates, productive government spending, and sustained economic growth are critical to achieving long-run labor market stability and reducing unemployment sustainably.

Heteroskedasticity Test

To test for heteroskedasticity in the residuals of the primary unemployment model (from which RESID^2 was derived), a Least Squares regression was performed with RESID^2 as the dependent variable and lagged unemployment, interest rates, and current fiscal variables as independent variables. This serves as a diagnostic test for the presence of non-constant variance in the error terms.

Table 4: Dependent Variable: RESID^2 Regression Results

The regression results for RESID^2 provide insights into the factors influencing the variance of the primary unemployment model’s residuals. The overall model has an R-squared of 0.584318, indicating that approximately 58.4% of the variation in the squared residuals is explained by the independent variables. However, the adjusted R-squared is lower at 0.298707, suggesting that some of the explanatory power might be due to the number of regressors rather than a strong underlying relationship. The F-statistic for the overall regression is 2.000000 with a probability of 0.144412, which is not statistically significant at conventional levels (e.g., 5% or 10%). This overall insignificance might suggest that, as a group, these variables do not collectively explain a significant portion of the residual variance.

However, examining individual coefficients reveals some interesting patterns:

UNEMPRATE(-2): The coefficient for unemployment rate lagged two periods is -0.002287 and is statistically significant at the 10% level (p-value = 0.0519). This indicates that past unemployment (specifically, two years prior) has a significant influence on the volatility of the model’s errors. A negative coefficient suggests that higher unemployment two periods ago is associated with lower residual variance, or vice versa.

IR(-2): The coefficient for interest rate lagged two periods is 0.001868 and is statistically significant at the 5% level (p-value = 0.0463). This implies that the interest rate two periods ago significantly affects the variance of the unemployment model’s errors. A positive coefficient suggests that higher interest rates two periods ago are associated with higher residual variance.

IR: The current interest rate (IR) also shows a marginal significance at the 10% level (p-value = 0.0836), with a coefficient of -0.001900. This suggests a contemporaneous inverse relationship with residual variance.

The significance of UNEMPRATE(-2) and IR(-2) in explaining the variance of the residuals suggests that the underlying unemployment model might exhibit heteroskedasticity, specifically conditional heteroskedasticity, where the variance of the errors depends on past values of unemployment and interest rates. This finding is crucial as it implies that the precision of the parameter estimates in the primary unemployment model might be compromised if not addressed.

Breusch-Godfrey Serial Correlation LM Test

The Breusch-Godfrey Serial Correlation LM Test was conducted to check for autocorrelation in the residuals of the primary model.

Table 4: Breusch-Godfrey Serial Correlation LM Test Results

The results of the Breusch-Godfrey Serial Correlation LM Test present a mixed picture. The F-statistic is 2.349330 with a p-value of 0.1762. Since this p-value is greater than 0.05, we fail to reject the null hypothesis of no serial correlation based on the F-statistic. This suggests that there is no significant first-order serial correlation in the residuals.

However, the ObsR-squared statistic is 7.597246 with a p-value of 0.0058. This p-value is highly significant (less than 0.01), which leads to the rejection of the null hypothesis of no serial correlation. The discrepancy between the F-statistic and ObsR-squared p-values can sometimes occur, especially in small samples. Given that the Obs*R-squared statistic is generally more reliable for larger samples, and its significance here, it suggests that there might be some form of serial correlation present, possibly of a higher order or a more complex structure, despite the F-statistic indicating otherwise for the specified lag (lag 1). This warrants further investigation or the use of robust standard errors in the primary unemployment model.

In summary, the unit root tests confirm the integration orders of the variables, making the ARDL approach suitable. The ARDL Bounds Test indicates a long-run relationship. The heteroskedasticity test suggests that the variance of the unemployment model’s residuals is influenced by lagged unemployment and interest rates, implying conditional heteroskedasticity. The serial correlation test provides conflicting results but leans towards the presence of autocorrelation, which would necessitate robust standard errors or appropriate model adjustments in the primary unemployment model.

CONCLUSION

This study embarked on an empirical investigation into the impact of fiscal and monetary policy on unemployment in Malaysia, utilizing annual time series data from 1995 to 2021. While the primary objective of the paper is to shed light on these relationships, the provided econometric results primarily focused on diagnostic tests of a hypothesized unemployment model. These diagnostics, however, offer crucial insights into the underlying statistical properties and reliability of such a model, which are foundational for drawing robust policy conclusions.

The initial phase of the analysis involved conducting Phillips-Perron (PP) unit root tests to ascertain the stationarity properties of the macroeconomic variables. The findings revealed that most variables, including unemployment rate (UNEMPRATE), GDP, interest rate (IR), government expenditure (GE), government tax revenue (GRTAX), government grants (GRGRANTS), and exchange rate (ER), were non-stationary at their levels but became stationary after first differencing. This indicates that these variables are integrated of order one, I(1). Broad money (BM) was found to be stationary at level, I(0). This mixed order of integration validated the appropriateness of employing the Autoregressive Distributed Lag (ARDL) bounds testing approach for cointegration.

The ARDL Bounds Test, designed to detect long-run equilibrium relationships among variables with mixed integration orders, yielded an F-statistic of 3.992644. While specific critical values were not provided, this value typically suggests the presence of a long-run cointegrating relationship at conventional significance levels. This implies that fiscal and monetary policy instruments, alongside unemployment and other macroeconomic variables, tend to move together in the long run, suggesting a stable equilibrium that the economy gravitates towards over time. The existence of such a long-run relationship is a prerequisite for understanding the sustained impact of policies.

A significant portion of the provided results pertained to diagnostic tests for model validity, particularly the regression of squared residuals (RESID2) to test for heteroskedasticity. This regression, which serves as a variant of the Breusch-Pagan or White test, aimed to determine if the variance of the errors from an underlying unemployment model was constant. The results from this test revealed that lagged unemployment (UNEMPRATE(-2)) and lagged interest rate (IR(-2)) significantly influenced the variance of the residuals. Specifically, UNEMPRATE(-2) had a statistically significant negative coefficient, implying that past unemployment levels are associated with changes in the volatility of the unemployment model’s errors. More notably, IR(-2) exhibited a statistically significant positive coefficient, indicating that past interest rate fluctuations contribute to the current volatility of the model’s errors. The current interest rate (IR) also showed marginal significance. Although the overall F-statistic for this heteroskedasticity regression was not significant, the individual significance of these lagged variables suggests the presence of conditional heteroskedasticity. This finding is crucial as it indicates that the precision of parameter estimates in the primary unemployment model may be compromised, and standard errors might be biased, potentially leading to incorrect inferences about the true impact of fiscal and monetary policies. Addressing this would typically involve using heteroskedasticity-consistent standard errors (White’s heteroskedasticity-consistent standard errors) or employing a GARCH-type model to explicitly model the time-varying variance.

Furthermore, the Breusch-Godfrey Serial Correlation LM Test was conducted to check for autocorrelation in the residuals. While the F-statistic suggested no significant serial correlation, the ObsR-squared statistic was highly significant (p-value = 0.0058), indicating the likely presence of autocorrelation. This conflicting result, particularly the strong significance of the ObsR-squared, suggests that the residuals of the underlying unemployment model might not be independently distributed. The presence of serial correlation can lead to inefficient parameter estimates and biased standard errors, necessitating the use of appropriate techniques such as ARMA error terms or Newey-West robust standard errors in the primary model estimation. 

Policy Implications

The diagnostic findings, while not directly quantifying the impact of policies on unemployment levels, provide critical insights for policymakers in Malaysia. The presence of conditional heteroskedasticity, particularly influenced by lagged unemployment and interest rates, implies that the effectiveness and predictability of fiscal and monetary policy interventions might vary over time depending on past economic conditions. Policymakers should be aware that the volatility of unemployment responses to policy changes is not constant. For instance, periods of high past unemployment or interest rate instability might lead to more volatile or less predictable outcomes from current policy adjustments. This calls for a more nuanced and adaptive approach to policy formulation, perhaps incorporating dynamic hedging strategies against unexpected fluctuations in the labor market.

The indication of a long-run relationship among the variables suggests that fiscal and monetary policies do have a sustained influence on the overall macroeconomic equilibrium, including unemployment. This reinforces the importance of consistent and well-coordinated long-term policy frameworks. However, the potential presence of serial correlation in the residuals underscores the need for robust econometric modeling in policy analysis to ensure that the estimated effects are reliable and efficient.

Limitations and Future Research

This study’s conclusions are constrained by the scope of the provided results, which predominantly focused on diagnostic tests rather than the direct estimation of fiscal and monetary policy coefficients on unemployment. A significant limitation is the absence of the primary unemployment regression model itself, making it challenging to directly quantify the magnitude and direction of the impact of each policy variable on unemployment.

Future research should aim to estimate a comprehensive ARDL model with unemployment as the dependent variable, explicitly incorporating the fiscal and monetary policy instruments identified. This would allow for the estimation of both short-run and long-run coefficients, providing a more direct understanding of their impact. Furthermore, addressing the identified heteroskedasticity through robust standard errors or GARCH modeling, and accounting for serial correlation, would enhance the statistical validity of such an analysis. Exploring structural breaks, non-linear relationships, and the inclusion of other relevant macroeconomic variables could also provide deeper insights into the complex dynamics of unemployment in Malaysia.

REFERENCES

1. Abd Majid, M. S. A., & Rasiah, R. (2014). Fiscal policy and unemployment in Malaysia: A vector autoregression (VAR) approach. Asian Academy of Management Journal of Accounting and Finance, 10(2), 57–73.
2. Arifin, N., & Ismail, M. (2023). Tackling youth unemployment in ASEAN: A policy integration framework. ASEAN Economic Bulletin, 40(1), 33–49. https://doi.org/10.1355/ae40-1d
3. Auerbach, A. J., & Gorodnichenko, Y. (2012). Measuring the output responses to fiscal policy. American Economic Journal: Economic Policy, 4(2), 1–27. https://doi.org/10.1257/pol.4.2.1
4. Bank Negara Malaysia. (2021). Annual report 2020. https://www.bnm.gov.my/documents/2021/20210331_ar2020_en_full.pdf
5. Fukao, K. (2020). Monetary easing and structural reform in Japan: Trends and debates.
6. Gechert, S., & Rannenberg, A. (2015). Fiscal multipliers in the OECD: A meta-analysis. Journal of Economic Surveys, 29(1), 143–172.
7. Ginting, E., & Aji, D. P. (2020). The role of fiscal policy in Southeast Asia’s economic development (Asian Development Bank Working Paper No. 612). Asian Development Bank.
8. He, J., & Liu, Y. (2022). Targeted monetary policy and employment: Evidence from China. Journal of Macroeconomics.
9. International Monetary Fund (IMF). (2022). Policy report.
10. Kuttner, K. N., & Posen, A. S. (2019). The great Abenomics experiment: Evaluating the impact of coordinated fiscal and monetary policy in Japan (Peterson Institute for International Economics Working Paper).
11. Mahmood, H., & Sadiq, S. (2020). Fiscal policy and unemployment in Nigeria: An ARDL approach. International Journal of Economics, Commerce and Management, 8(1), 1–15.
12. Mishkin, F. S. (2019). The economics of money, banking and financial markets (13th ed.). Pearson Education.
13. Monastiriotis, V., & Antoniou, F. (2013). The unemployment effects of fiscal shocks in Greece. Applied Economics, 45(21), 3121–3134.
14. Okun, A. M. (1962). Potential GNP: Its measurement and significance. Proceedings of the Business and Economic Statistics Section, American Statistical Association, 98–103.
15. Onwuka, C. E. (2022). The impact of fiscal and monetary policy on unemployment rate in Nigeria (1981–2020). Journal of Economic Research & Reviews, 2(3), 226–235.
16. Pesaran, M. H., & Shin, Y. (1999). An autoregressive distributed-lag modelling approach to cointegration analysis. In S. Strom (Ed.), Econometrics and economic theory in the 20th century: The Ragnar Frisch Centennial Symposium (pp. 371–413). Cambridge University Press.
17. Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326. https://doi.org/10.1002/jae.616
18. Pham, H. T. (2022). Monetary policy and unemployment in ASEAN: Evidence from panel VAR. Journal of Asian Economics, 78, https://doi.org/10.1016/j.asieco.2022.101456
19. Saari, M. Y., Ahmad, N., & Abdullah, A. (2018). The impact of government spending on unemployment in Malaysia: A CGE approach. International Journal of Economics and Management, 12(1), 13–28.
20. Stockhammer, E., & Sturn, S. (2011). The impact of monetary policy on unemployment hysteresis. Applied Economics, 44(21), 2743–2756.
21. Turrini, A. (2013). Fiscal consolidations and unemployment: A panel analysis (European Commission Discussion Paper).