Introduction

Wagner’s law is a principle named after German economist Adolph Wagner (1958), who established the law of expanding state expenditure. This law emphasizes the importance of government spending as an endogenous variable that tends to grow with other economic aggregates to feature a “progressive state” (Wagner, 1958). The law also postulates that government spending is an outcome of economic growth and has a causal relationship that moves in this single direction. In contrast, Keynes (1937) identified government spending as an exogenous factor that stimulates economic growth, viewing the reverse relationship between public spending and economic growth.

A preponderance of studies shows great interest to both academics and policymakers in the aforementioned nexus. However, a new trend in economic analysis emerged that attempts to view government spending not only as an aggregate but also as an aggregate with different spending components. The argument made was that this latter view adds to the holistic view of the relationship between economic growth and government spending (Dritsakis and Adamopoulos, 2004; Gupta, 1969; Hondroyiannis and Papapetrou, 1995). Similarly, political economic arguments that focused precisely on the impact of military spending on economic growth contributed to the same investigation into whether this impact is exogenous or endogenous to the growth model (Abdelfattah et al., 2014).

Egypt, for instance, according to the World Bank (2021), has experienced significant changes in military spending. It reached its highest point between 1967 and 1973 due to the Arab–Israeli war, then fell from 56 to 14% between 1974 and 1981. In 1991, the U.S. released $7 billion in military debts on Egypt. Finally, military spending reached its peak in 2011 during the political uprising (Abdelfattah et al., 2014), and continued in 2012 to support the war against terrorism and secure the Egyptian borders with Gaza and Libya.

Furthermore, to add analytical depth, both education and health components of human development are included to determine the dynamic causal relationship between them, military spending, and economic growth. The assumption is that Egypt, which has undergone different stages of political conflict, had a preference for military spending that negatively impacted economic and human development over the past decades. The share of education in government spending was at its maximum in 2010, at 3.3%, but experienced a significant drop from 2.2% in 2011 to 1.7% in 2014, which was a period of political transition and unrest. Afterward, education expenditure started to increase, reaching 3.2% in 2017. As for government spending on health, Egypt has witnessed a low budget share for health compared to other spending components, averaging 3.7% of GDP. The Ministry of Health and Population has increased government spending on health from 2.2% in 1996 to 3.3% in 2001 (Gericke et al., 2018). Similarly, the share of health in government expenditure dropped from a peak of 2.3% in 2010 to 1.7% in 2016, and then increased to 3.2% in 2017.

The existing literature on the relationship between military spending and economic growth in Egypt has primarily focused on examining the impact of military spending on political instability and overall economic growth (Maher and Zhao, 2022). However, these studies have not thoroughly investigated the relationship between military spending and specific components of government spending, such as education and health. Additionally, they have not explored the reverse relationship between military spending and health and education expenditures. For instance, Maher and Zhao (2022) did not allocate their variables to budgetary spending components, limiting their analysis.

In contrast, Lin et al. (2015) conducted a comprehensive study across 29 member countries of the Organization for Economic Cooperation and Development (OECD) to investigate the trade-off between military spending and social welfare spending, specifically health and education expenditures. Their findings suggested that as military spending increased in these countries, there was also an increase in health and education spending, supporting the Keynesian view of military spending. This suggests that military spending may promote economic growth and enable governments to allocate more resources to these crucial areas. However, this study was not specific to Egypt and did not directly examine the relationship between military spending and economic growth in the country.

Furthermore, previous research conducted by (Ali, 2011) on Egypt’s military spending and economic growth found inconclusive evidence to support his hypothesis. Similarly, Eldemerdash and Ahmed (2019) investigated the alternating views of Wagner and Keynes on the effect of military spending on Egyptian economic growth but focused solely on the overall government spending-to-GDP ratio without delving into the specific impact on health and education expenditures. While their contributions align with other empirical studies, they did not detect causality from government spending to GDP, supporting Wagner’s main hypothesis (Ghazy et al., 2021; Abu-Bader and Abu-Qarn, 2003).

This research contribution is innovative in its attempt to narrow the research gap in studies investigating the effect of government spending on economic growth. First, one contribution is investigating this relationship with respect to military spending using two different theoretical frameworks (Wagner and Keynes), unlike most studies that adopt the Keynesian school called Military Keynesians (Atesoglu, 2002; Dunne, 2012; Hossein-zadeh, 2009; Nordhaus, 2002). Second, this study analyzes the effect of military spending on economic growth and investigates whether military spending has a crowding-out effect on other aspects of human development contributing to economic growth, namely health and education spending.

The remainder of this paper is organized as follows. The first section introduces this topic. The second section presents a literature review on the theoretical framework and empirical studies. The third section provides research methodology. The fourth section presents a summary of the empirical results, and section five discusses the model results. The final section concludes the paper.

Literature review

Theoretical background

Wagner (1958) asserts a direct causal relationship between economic growth and government spending in the same direction. The first formulation was introduced by (Goffman and Mahar, 1971; Oates, 2005; Peacock and Wiseman, 1961). Quantified extensions of Wagner’s law were subsequently introduced and summarized by (Akitoby et al., 2006; Kibara Manyeki and Kotosz, 2017; Uppal and Glazer, 2015).

In a different vein, Keynes (1963) stated that market economies could not generate full employment; therefore, government expenditures are a catalyst for boosting aggregate demand, which determines the amount of output and income. This view was supported by by Barro (1990), who stated that government expenditure also affects the production function through an endogenous growth hypothesis.

The Keynesian view was further developed to establish the “Military Keynesians” line of political economic thought that views the impact of military spending as stimulating aggregate demand, investment, and employment (Atesoglu, 2002; Nordhaus, 2002). Alternatively, other scholarly arguments do not support this positive effect. A study of the U.S. economy by Melman (1978) asserted the damaging effect of increasing military spending by creating a lack of competitiveness, increasing bureaucracy, investment disincentives, externalities, and the negative spillover effects directed from the military to the civil sector. Similarly, (Dunne, 2012; Hossein-zadeh, 2009) argued that the increase in government spending allocation toward military spending in the United States has crowded out both physical and human investments, which, in turn, negatively affect economic growth. Finally, (Dunne, 2012; Gold, 2005; Kinsella, 1990; Payne and Ross, 1992) argue that military spending has a negative or insignificant effect on economic growth, although not in the context of the Keynesian spending model. Studying Wagner’s Law versus the Keynesian view was not isolated from inspecting the causal relationship between growth and different components of government spending, including military spending. The argument made was that empirical studies were deficient in that they only viewed government spending in aggregate, which does not capture the effect of spending on development sectors, such as health, education, administration, and other aspects of development, without analyzing how spending is allocated (Dritsakis and Adamopoulos, 2004; Gupta, 1969; Hondroyiannis and Papapetrou, 1995).

Samudram et al. (2009) used the ARDL model and bounded tests for the Malaysian economy to provide evidence of a long-term relationship between total expenditures allocated to military spending, education, development, and agriculture and gross national product (GNP). Their structural break analysis at a cut-off point in 1998 revealed that this long-term causality is bi-directional between GNP and spending on the administration and health sectors; therefore, it supports both Wagner’s law and the Keynesian view. In contrast, only Wagner’s Law was supported by the unidirectional causality relationship between the GNP and other spending allocations, including military spending.

Abu-Bader and Abu-Qarn (2003) was in the same vein as studies that attempted to test the validity of Wagner’s law in Egypt in relation to military spending; they tested the causal relationship between economic growth and civilian versus military spending. Their focus was on countries involved in Arab–Israeli military conflict, including Egypt. However, this study differs from the current study regarding the desegregation of civilian expenditures into different components affecting human capital, namely education and health. This study provides a comparative analysis of these components in relation to military spending. It also investigates the directional relationship between military spending and spending on health and education to determine whether the allocation of resources between these sectors affected human capital, a major catalyst for growth and development. Moreover, their study provides additional insights into the share of military spending that seems minor relative to the share of health and education in public spending, and that can have a long-term impact on economic growth.

Abdelfattah et al. (2014) asserted that most existing studies adopted the Keynesian and classical approach to the growth expenditure model, as shown from their literature survey on the effect of military spending on economic growth—unlike the present study, which analyzes this relationship that tests Wagner’s law and the Keynesian hypothesis. Therefore, this study fills a research gap by incorporating Wagner’s law and disaggregated spending into the analysis.

Empirical review

The Keynesian view of military spending has three main empirical applications, as summarized in (Dunne, 2013). The first applies Granger causality methods that conduct simple bivariate relations between military spending and growth and examines the long-run relationship between both variables through a cointegration analysis and vector autoregression model (Dunne and Vougas, 1999). The second considers structural models established using the original Keynesian IS–LM model (Atesoglu, 2002; Pieroni et al., 2008; Smith and Tuttle, 2008). These studies used the log of real GDP, military spending, civil spending, and real interest rate, and their results suggested a long-term relationship between GDP growth and military spending. The third, by Dunne and Nikolaidou (2005) estimate Keynesian models using the aggregate production function to capture effective demand, for which output was regressed as a function of military spending, in addition to the share of non-military spending to GDP and the share of investment in GDP as independent variables. This method was applied to three EU countries, Greece, Portugal, and Spain, for which the direction of causality from military spending to output was evident only in Greece, whereas the reverse of causality from output to military spending was evident for the three countries. With respect to the other variables used in the model, no conclusive growth showed that a causality relationship exists between military and non-military spending and investment as a share of GDP.

Numerous empirical studies on Wagner’s law for aggregate government spending using the ARDL model for different countries proved its validity (Atilgan et al., 2017; Burney, 2002; Islam, 2001; Kesavarajah, 2012; Magazzino, 2012; Manamperi, 2016; Narayan et al., 2008; Samudram et al., 2009). Another group with the same research objective used a cross-country analysis to find evidence of Wagner’s law, as in (Karagianni and Pempetzoglou, 2009; Shelton, 2007).

A third group investigated Wagner’s law and the Keynesian view to test whether they are valid for other countries’ economies (Ghazy et al., 2021; Paparas et al., 2019).

More recently, various studies have explored the crowding-out effect of military spending on different aspects of government expenditure. Jesmy et al. (2015) investigated the impact of military expenditure on education in five South Asian countries from 1980 to 2013, using panel regression methods. They found a negative relationship between military spending and the quality of education in these countries, which supports the main hypothesis of this study.

Azam (2020) examined the effect of military spending on economic growth in 35 non-OECD countries from 1988 to 2019. Using a multivariate regression model with the augmented production function, he found a negative impact of military spending on economic growth. His causality test results also revealed a bi-directional causality between military expenses and economic growth.

Inal et al. (2022) used panel cointegration tests to determine the long-run relationship between military expenditures, economic growth, innovation, and labor productivity in countries characterized by Startextensive militarization. Their study concluded that there is a long-run and causal relationship between military spending and the utilized variables.

Ikegami and Wang (2023) investigated the crowding-out effect of military spending on the health sector in 116 countries from 2000 to 2017. They found that military expenditure had a positive impact on the demand for healthcare but also a significant crowding-out effect on health spending. This result aligns with the current study’s findings. On the contrary, Biscione and Caruso (2021) and Coutts et al. (2019) found no significant impact of military spending on health expenditure in countries with transition economies and countries in the Middle East and North Africa (MENA) region, respectively.

Becker and Dunne (2023) used data from NATO and the EU on the decomposed items of military spending for 34 countries over 49 years to identify which component of military spending negatively affected economic growth. Their study concluded that military personnel spending had a more significant impact than other military operational spending. However, more recent study by Raifu and Aminu (2023) investigated the effect of military spending on economic growth in MENA countries from 1981 to 2019. They used quantile panel regression via moments and found that military spending positively affects economic growth, confirming the Keynesian theory.

Wang and Su (2021) used the mixed frequency vector autoregression model to find a direct causal effect of crude oil dependence on military expenditure. In another study by Wang et al. (2021) explored a different causal relationship between crude oil dependence, CO2 emissions, and military expenditure for oil-importing countries. They employed the bootstrap autoregressive distributed lagged model with a Fourier function and found a cointegration between the three variables in China, Italy, and India.

Eventually, the economic effects of government spending on economic growth and military spending have been extensively studied from various perspectives. However, the literature lacks a consensus on whether military spending aligns with Wagner’s law, Keynes’ law, or both, in terms of its effects on economic growth. Additionally, there has been no previous investigation into how much military spending affects the budget allocation of spending on health and education, and the direction of this impact in the short and long terms. Furthermore, there is a significant gap in research on the nexus between military spending and economic development in Egypt. Hence, this study contributes to the existing literature by specifically examining the directional impact of military spending on other types of developmental spending. The following sections will discuss the empirical methodology employed and the data used for the study.

Methodology

Data collection and transformation

This study was conducted on the Egyptian economy from 1980 to 2021. Secondary data samples were retrieved from the World Bank Indicators, UNICEF, SIPRI Military Expenditure Database, UNESCO, and Ministry of Finance of Egypt. Some variables for health and education were missing data and were estimated using the interpolation method. The study primarily uses the Pryor (1968) version of Wagner’s law and follows, in particular, the research methodology of Al Qudah et al. (2020). A theoretical framework of the general macroeconomic model is proposed that considers the interdependence of economic growth, military spending, and government spending on health and education. All variables used in this study were expressed in natural logarithmic form.

One of the main issues in economics is that the long-run relationship must be estimated for policymaking. As a result, when utilizing variables that are not co-integrated, we would potentially fail to effectively estimate the long-run relationships and their impacts. The main purpose of adopting the ARDL technique is to use variables with different integration orders and/or variables that might suffer from unit root problems to a certain extent. Owing to the non-stationarity problem, the transformation of the difference is applied along with the logarithmic transformation. As a result of that, the mixture between ARDL and the error-correction model (ECM) technique is employed to capture both long-run and short-run effects by estimating the equilibrium coefficient (Chandio et al., 2019).

Empirical framework

First, summary statistics and correlation analysis were used to demonstrate the relations between the variables and to detect possible multicollinearity, followed by the variance inflation factor (VIF) test to rule out any presence of multicollinearity. It should be noted that if the variable has VIF higher than the value of 10, it will be considered to cause the problem of multicollinearity and could be excluded (Alin, 2010). Two unit root tests were chosen—Augmented Dickey Fuller (ADF) and Phillips–Perron (PP)—to determine whether the data are stationary at level or at first difference or both to proceed to the cointegration test by the long-term relationship detection between variables, followed by the Chaw test for a structural break to determine whether the time series shows any time breaks from 1980 to 2021, which can lead to major forecasting errors and general model unreliability. The last step before testing for cointegration is to determine the optimal lag selection tests to choose the optimal lags to run the models. Afterward, a cointegration test for Pesaran et al. (2001) on the long-term relation between variables was employed. To add more depth to the analysis, the Granger causality test was used to determine the short-term interchangeable relationship between military government spending and government spending on both health and education. Finally, to capture both short- and long-term effects, the ECM within the ARDL was applied based on the direction of causality identified in the Granger causality test.

Empircal analysis

Summary statistics and correlations

Table 1 presents a brief descriptive analysis of these correlations. The descriptive analysis shows the distribution properties of the individual variables and the correlation matrix shows the relationship between these variables in our proposed model.

Table 1 Summary statistics and correlations.
Full size table

According to Table 1, the mean of Egypt’s GDP is $156 billion, with a standard deviation of $64.2 billion, which is very high, indicating that the data deviate from the mean, as shown by the minimum value of $24.8 billion and a maximum value of $52.6 billion. The maximum occurred in 1982 because of the open-door policy implemented in Egypt, which significantly increased the national income. The lowest income occurred in 2011, when Egypt faced a revolution that significantly affected its economy.

The mean of government expenditures is $26.7 billion, with a low standard deviation of $16.4 billion that indicates a small gap between the minimum and maximum values of the government expenditure variable. The minimum is $41.4 billion, and the maximum is $47.2 billion. The minimum in 1991 was the result of a reduction in government revenues after the oil price crash during 1985–1986.

A comparison of the different government expenditures for military, health, and education showed that the lowest mean was for the military, at $4.22 billion, with a standard deviation of $1.23 billion. The highest mean was for education, at $14.56 billion, with a standard deviation of $4.31 billion. A high standard deviation indicates a significant gap between the minimum and maximum values of education expenditure of $6.23 billion, and the maximum was $19.86 billion. The small standard deviation indicates that the minimum and maximum values of military spending were $1.5 billion and $5.05 billion. Health expenditures were in the middle, with a mean of $6.73 billion and a standard deviation of $2.20 billion, similar to the military, with minimum and maximum variables of $4.05 and $8.07 billion, respectively. The reason for the wide gap between the minimum and maximum military spending variables is that the data used started in 1980, during a period of peace and stability. Subsequently, the 2011 Revolution and various political administrations prompted an increase in military spending.

The correlation matrix in Table 1 shows the correlations between GDP, military spending, and government expenditure on health and education. A negative, strong, linear relationship exists between GDP and military expenditure, which aligns with the theory, given that classical theory states that an increase in military expenditure negatively affects economic growth because it lowers private investments and domestic savings. This effect leads to lower consumption due to lower aggregate demand because higher military spending increases interest rates and crowds out private investments.

A weak positive relationship exists regarding the correlation between GDP and government expenditure variables on health, as stated in Romer’s growth theory theory (Romer, 1994; van Zon and Muysken, 2001), because health is an important factor that affects labor productivity and human capital accumulation. The correlation between GDP and government expenditure on education indicates that they are strongly, positively, and linearly related because government expenditure increases economic growth through knowledge, innovation, research and development (R&D), and efficiency in human capital, as stated by classical theory. For all values, the p-value was less than significant (0.05); therefore, a linear relationship existed between them.

Variance inflation vector test

As shown in Table 2, that all VIF values are less than 10 indicate the absence of multicollinearity between the tested independent variables and the dependent variable of LGDP.

Table 2 VIF test.
Full size table

Unit root test

As discussed in the methodology section, checking the stationarity of variables is important in the time series to proceed with further testing and estimation techniques. Table 3 provides the results of an ADF test of the unit root tests conducted for Egypt for the period 1980–2021.

Table 3 Unit root tests.
Full size table

The null hypothesis of both the ADF and PP unit root tests assumes that the series is nonstationary at the level. Table 3 shows that without differentiating the log variables, the null hypothesis of having unit roots cannot be rejected at 1, 5, or 10% significance for LGDP, LMIL, LGOVHEL, and LGOVEDU, respectively. However, when the first difference is obtained for the variables in both the ADF and PP tests, the null hypothesis of non-stationarity is rejected at the 5% significance level for LGDP and 10% significance level for LMIL, LGOVHEL, and LGOVEDU.

Structural breaks

Exogenous shocks in an economy may have a permanent and immediate impact on many economic variables. Therefore, testing for structural breaks is crucial for avoiding unreliable results. The Chow test is used to examine the structural breaks in the given time series for the Egyptian economy during 1980–2021, with the dependent variables being LGDP, LGOVMIL, and LGOVHEL LGOVEDU.

As observed in Table 4, the structural break results verify the results of the unit root tests because all variables are stationary at level. The structural breaks selected by the tests at this level were 1990 and 2011 for the variables tested. This finding is understandable because government spending was significantly affected and reached its highest level in 1990 as a result of the Economic Reform and Structural Adjustment Program (ERSAP), with the main objective of loosening price restrictions, encouraging private investments, and easing industrial investment procedures (Korayem, 1997). In 2011, Egypt’s revolution, government spending, and GDP declined significantly, and the real GDP growth rate reached a dreadful of 1.78%. To solve this problem, an expansionary fiscal policy was implemented after the revolution to address political unrest (Khan and Miller, 2016).

Table 4 Structural break using Chow test.
Full size table

Optimal lag selection

To determine the optimal number of lags to be utilized in running the cointegration test and, further, the model, several criteria are estimated for different numbers of lags. The optimal number of lags of one for the model was selected by examining the optimal lag length chosen from the different information criteria in Table 5. The optimum number of lags agreed upon is one because it has the highest statistical significance. Furthermore, the same lag length is used to subsequently test the ECM and ARDL models to avoid a lack in the number of observations and loss in degrees of freedom.

Table 5 Lag selection criteria.
Full size table

Pesaran–Smith cointegration test

To test whether a long-term relationship exists among government spending, military spending, government expenditures on health, government expenditures on education, and GDP, a Pesaran–Smith cointegration test is used. This test assumes that the null hypothesis is that no relationship or cointegration exists for a certain number of lags.

The test shows no cointegration at level, given that the test statistic is −7.3279 with a p-value of 0.9827, reflecting the non-rejection of the null hypothesis in the presence of no cointegration. After taking the first difference for the logged variables, the test statistic is −28.41 with a p-value of 0.0000, indicating that the covariates and dependent variable are co-integrated after the first difference.

Discussion

Granger causality model discussion

The Granger causality test is used to indicate whether the variables used in the model Granger cause each other in the short-term. The test was estimated to last 38 years, as two lags were taken, as suggested by the optimum lag number in Table 6. The null hypothesis for the Granger causality test is that GDP can Granger-cause LMIL, LGOVHEL, and LGOVEDU, and vice versa. In addition, whether the same independent variables Granger cause each other in terms of military spending might negatively affect health and education spending, leading to deterring economic development, as shown in Table 6.

Table 6 Granger causality test (GS components with GDP).
Full size table

A causality test was conducted to identify the direction of the relationships among the study variables—GDP and government spending components of military (LGOVMIL), health (LGOVHEL), and education (LGOVEDU) expenditures.

The results indicate a unidirectional relationship between the government spending components and GDP at the 5% level, thereby supporting the Keynesian view that government spending affects GDP. In contrast, causality is not significant from the components to GDP at all significance levels, indicating that Wagner’s law is not applicable to the Egyptian economy. However, the direction of the relationship in the short and long terms, whether positive or negative, is yet to be revealed when both ARDL and ECM models are employed. This supports the empirical results in Ghazy et al. (2021).

In addition, military expenditure Granger causes both health and education spending, as indicated by a significant p-value, which indicates rejection of the null hypothesis of no relationship between them in that direction. By contrast, health and education do not Granger-cause military spending. The results confirm the empirical findings of Inal et al. (2022) and Ikegami and Wang (2023).

The correlation analysis in Table 1 shows that the coefficient of correlation between military spending and spending on health and education is –0.45 and 0.91, respectively. This finding implies that increasing spending in the military sector negatively affects health but positively affects education. This phenomenon might be the result of the presence of military elementary, secondary, and higher education in Egypt, which may have a significant share of military spending. The lack of information and data on military budget allocation is a limitation; the availability of such data would make it an area for further research on how spending on military education can affect Egypt’s education sector.

ARDL model discussion

The ARDL with ECM technique was used to assess the dynamic relationship between the set of independent variables and their impact on real GDP. ECM was used to evaluate both the long- and short-term partial effects of the independent variables on the dependent variables, as shown in Table 7.

Table 7 ARDL integrated with ECM test.
Full size table

The coefficient of the lagged residuals shows that the average adjustment speed to equilibrium is 0.36%, holding the other variables constant. Regarding the short-term shown in Table 7, when military spending increases by 1%, GDP is expected to decrease by an average of 0.09%, holding other variables constant, as stated by the theory. Because the p-value was less than the significance level (0.01), this parameter was significant. In the long-term, when military expenditures increase by 1%, GDP is expected to decrease by an average of 6.98%, holding other variables constant, which is a greater burden that the government spills on future generations.

When government expenditure on health increases by 1% in the short-term, the GDP is expected to decrease by an average of 0.06%. However, long-term GDP is expected to increase by 2.43%, confirming the Romer (1994) endogenous growth model on human capital formation, which suggests further empirical studies van Zon and Muysken (2001) who found that a sharp decline in economic growth is evident for countries with high rates of health deterioration and poor health sectors, leading to low productivity; hence, the long-run positive association between health spending and economic growth supports this empirical evidence. The p-values for both the long and short terms were less than the significance level. Therefore, these parameters are significant for this variable.

When government expenditure on education increases by 1%, the average GDP is expected to decrease by 0.045% in the short-term because time is needed for education necessary for developing human capital to influence economic growth. However, the coefficient increases by 5.33% over the long-term. The p-values for both the long and short-term were less than the significance levels. Therefore, the parameters are significant for this variable.

The R2 value for the model as a whole show that the independent variables explain 59.63% of the variation in the dependent variable, while the rest is the result of errors.

Conclusion

This study examines the validity of Wagner’s Law versus the Keynesian view of the Egyptian economy from 1980 to 2021. The components of the relationship between government spending and GDP growth were analyzed in terms of the direction of causality and their long- and short-term relationships.

The Granger causality test revealed a unidirectional relationship between government spending and economic growth components, rejecting Wagner’s law and supporting both the main Keynesian and military Keynesian views regarding the presence of the effect of government spending on GDP. In addition, evidence exists for a causal relationship between military spending and health and education spending, and the positive association between military spending and education spending suggests areas for further research, provided that the limitation resulting from unavailable data on the spending channels to which military education is directed is resolved. Moreover, the ARDL and ECM for the disaggregation of government spending in Egypt show that military spending has a negative effect on GDP in both the short and long terms. However, government spending’s effect on GDP has a negative effect in the short-term on health and education but a positive effect in the long-term.

These results raise important policy implications that suggest that the orientation of government spending to enhance economic growth should be revisited. Military spending must decrease because it has a negative impact on economic growth in both the short and the long-term. In addition, government spending on health and education should be increased, because the effect seems negative in the short-term. However, the long-term effect is positive and provides significant benefits to economic growth through innovation, productivity, R&D, and technology. Egypt will be better off with lower government spending because increases in government spending are mainly used as a countercyclical tool to smooth the country’s business cycles.

Further analysis of the effect of government spending on infrastructure should be incorporated into different models to test this relationship, because it is a major contributor to economic growth. Regarding recommendations for future research, comparing Egypt to another emerging country that surpassed it through economic growth, such as South Africa, would provide valuable policy recommendations and add an infrastructure component to assist in fully capturing government spending.