The efficiency of government finanical expenditures before and during the COVID-19 pandemic: A cross-country investigation

Abstract

This paper examines the way government spending affected productivity and its decomposition before and during the COVID-19 outbreak. Using panel data from 158 economies, the research shows that spending on health care increases productivity, while spending on the military slows down productivity and its decompositions. These effects are even greater in the context of COVID-19, showing that spending on health care and avoiding conflict and military escalation will be important for future economies to grow in a sustainable way.

1. Introduction

Common arguments show that government spending plays an important role in boosting productivity and reducing the social impact of pandemics like COVID-19 (Anwar, 2001; Goodell, 2020; Makin and Layton, 2021; Ruiz, 2018). One positive effect of government spending on economic outcomes is to cope with market failures. This means that government spending (i) improves market equilibrium by providing more public goods and goods with positive externalities (like the COVID-19 vaccine); (ii) limits asymmetric information by expenditure on product quality, market management, and insurance bureaus; and (iii) prevents people from acting irrationally in the economy (e.g., refusing to wear masks during the peak of the COVID-19 pandemic). Zhao et al. (2021) show that long-term productivity growth is affected by how much the government spends on public goods that have a high return for the community but not for private businesses.

Keynesian economists show that government spending increases aggregate demand in the economy, reduces temporary unemployment, and makes it easier to invest in public goods that improve social security. This, in turn, improves the economy's productivity and well-being. For example, post-Keynesian scholars say that whether the government should spend more on the economy depends on how the economy is doing at the moment (for example, whether it follows a profit-led demand regime or a wage-led demand regime) (Parui, 2021). A rise in government spending also has an indirect effect on productivity by reducing inequality. High inequality hurts the economy by (i) wasting human capital and (ii) limiting talent and opportunities for people on low incomes.

In contrast, increasing government spending can send a misleading signal to the economy and result in consequences that are hard to predict. Neoclassical and Austrian economists maintain that when the government spends heavily on items that are not essential (such as defense and primary education), the results are often unexpectedly bad (Rothbard, 2009). For example, taxes impose a gap between the gross and net return on savings, and this in turn slows the accumulation of capital, causes uncertainty, and hurts long-term economic growth (Cavallari and Romano, 2017). To give another example, government spending makes it harder for the private sector to invest and hire employees. Thus, if government stimulates production, this will require more labor and investment, pushing up real wages and interest rates. The result will be less work and investment in the private sector (Dinh and Canh, 2019). Raising taxes will also make market incentives less attractive, impeding economic growth (Qian and Weingast, 1997).

The institutional sclerosis theory says that when governments try too hard to get what they want, it hurts the market function of the economy (Olson, 2008). This is especially true in countries that are in transition, like Vietnam and China. In corrupt countries, the problem of rent-seeking and policies that do not aim for the good of the whole country, but only advance the interests of a small group, make it difficult to improve the efficiency of the economy through more government spending. In fact, corruption can cost developing countries between 0.3% and 1.9% of their GDP (Olken and Pande, 2011), a problem that may get worse if there are no clear ways for governments to keep track of it (Olken and Pande, 2011; Van Vu et al., 2018).

The mixed findings concerning government spending on productivity comes from several sources. First, the link between government spending and productivity depends on other catalysts, such as global trade, economic development levels, government effectiveness, and contextual factors (Romer and Romer, 2010). More importantly, most previous studies regarded TFP as an aggregated index, making it hard to investigate the details of the relationship between government spending and productivity. Given the arguments above, this study makes several contributions, as follows. First, it provides a fresh application of the Stochastic Frontier Analysis-FP (SFA-FP) index, evaluating TFP change and its components at the international level. Second, our study assesses the impact of different types of government spending on TFP and its components before and during COVID-19. Thus, the differing effects of various types of government spending on TFP components may help us understand how government intervention affects the performance of the economy and make sense of the contradictory findings in the literature.

2. Research context

The world's average annual TFP growth from 2014 to 2019 was only about 1% (see Appendix 1). In 2020, affected by the COVID-19 pandemic, the TFPI decreased by about 2.2% and resulted in additional consequences for global production (e.g., supply chain disruptions and shifting economic structures). The decline of TFPI derives mainly from OSEI (output-oriented scale efficiency index), implying that the use of resources (i.e., labor, capital) is inefficient in scale and allocation. Specifically, OSEI has continuously decreased from 2014 to 2019 at a rate of about 0.6–0.8% per year. This finding also implies the necessity to enhance the role of efficient resource allocation through international exchange, as foreseen by neoclassical economists. Inefficiencies of scale and allocation can be clearly illustrated through the 1929 overproduction crisis (excessively optimal scale) and the Venezuelan food crisis (lack of resource allocation by manufacturing industries). Thus, improving OSEI is the key to TFPI growth in the coming period. Furthermore, changes in the institutional environment, reflected by OEI (output-oriented environment index), also reduced TFPI, although this decline was slight throughout the study period.

Under pressure from the COVID-19 pandemic, OSEI improved in 2020, showing 1% growth. This phenomenon can be explained by the limited supply/demand pressures imposed by the pandemic, so that organizations allocated resources more efficiently. Even so, COVID ravaged economies through various channels, such as global supply chain and labor and capital market disruptions, reflecting the SNI component. The results shown in Appendix 1 confirm that the 4.3% decrease in SNI components was the main cause of the erosion of TFP.

Changes in TFP and its decompositions in 158 countries from 2014 to 2020 can be seen in Fig. 1, Fig. 2 . The annual TFP growth of China (4–5%/year), Vietnam (3%/year), and a few other countries (e.g., Somalia) are significant, signs of possible success in future economic transformation. It is also important to note that the larger the economy (e.g., the United States and the United Kingdom), the slower the economic growth rate, as predicted by endogenous growth theory. Notably, the similarity in color diffusion of the TFP figure and its component figures shows differences in the component contribution to TFP growth. For example, although China and Vietnam have a TFP growth rate of about 3–5%, OSEI's contribution is relatively limited (showing even negative change).

Fig. 1.

Annual TFP growth rate (%), 2014–2020 on average.

Fig. 2.

Annual TFP and its component growth rate (%).

Source: Source: Authors. Note white shows no data available.

3. Methodology and data

3.1. Data sources

The study uses three main sources of information. The first source is data from the World Bank about the inputs and outputs of the production process (labor, total capital, gross domestic product [GDP]) and some control variables for the second step (trade size, total land, quality of government) for 158 countries from 2014 to 2020. The World Governance Indicator (WGI), which looks at six aspects of the quality of national institutions, was the second source. Thus, institutions are very important for the long-term growth of economies, especially in countries that are in transition, where balancing the roles of the private and public sectors is key. The third source is the Mobile Connective Index (MCI), which shows how well and how equally electronic devices and technologies in the digital age can connect to each other (e.g., 3G and 4G connectivity).

3.2. Methodology

Given the level of technological production in the period specified (2014 to 2020), the frontier of the period- and environment-specific production possibilities with vector input, output, and environment x,  y,  and z (respectively) set by Tt(zit) [(x, q): x] can produce q in period t in an environment characterized by zit. The TFP of country i in period t can be defined as:

$$TF{P}_{it}=\frac{Q\left(q_{it}\right)}{X\left(x_{it}\right)}$$ (1)

where, q_it denotes the vector of L outputs and x_it represents the vector of M inputs. Q(.) and X(.) are scalar aggregator functions with non-negative, non-decreasing, linear, and homogeneous attributes.

The changes in TFP are decomposed into several components (Minh and Quang, 2022; O'Donnell, 2018, 2016) in the output-oriented approach, including (i) output-oriented technical efficiency (OTE); (ii) output-oriented scale mix efficiency (OSEI); (iii) environmental efficiency (OEI); and (iv) statistical noise (SN). The index TFPI considers the change between the TFP of a country i in year t and the TFP of a country k in year s, and this change can then be computed as the ratio of TFP_it over TFP_ks as below.

$$TFP{I}_{ksit}=\frac{D_O^{\overline{t}}\left(\overline{x},q_{it},\overline{z}\right)}{D_O^{\overline{t}}\left(\overline{x},q_{ks},\overline{z}\right)}\frac{D_I^{\overline{t}}\left(x_{ks},\overline{q},\overline{z}\right)}{D_I^{\overline{t}}\left(x_{it},\overline{q},\overline{z}\right)}$$ (2)

In our application, we consider only the economic factor, represented by gross domestic product (GDP). The index presented above is then simplified in:

$$TFP{I}_{ksit}=\frac{q_{lit}}{q_{lks}}\prod _{m=1}^M{\left(\frac{x_{mks}}{x_{mit}}\right)}^{\frac{{\beta }_m}{r}}$$ (3)

The SFA method, which is used here, allows the estimation of β₁,β₂,…, β_m coefficients. In this case, we consider a sample of N countries followed during T years, for which the study observed the inputs, J* environmental factors, and outputs of the production process. The distance function defined is:

$$ln{q}_{it}={\phi }_i+\lambda t+\sum _{j=1}^{J^{*}}{\delta }_{j}ln{z}_{ijt}+\sum _{m=1}^M{\beta }_{m}ln{x}_{imt}+u_{it}+v_{it}$$ (4)

where, u_it = $-Ln{D}_O^t$(x_it,q_it,z_it) is non-negative technical efficiency, commonly known as learning by doing capacity. The error term v_it accounts for functional error and omitted variables, which can occur because (i) the distance function is not a Cobb-Douglas function and (ii) because of other sources of statistical and stochastic noise (e.g., measurement errors). λ indicates unobservable conditions changing over time that allow for the rate of technical change. $\sum _{m=1}^{M}ln{x}_{imt}$ and $\sum _{j=1}^{J^{*}}ln{z}_{ijt}$ are conventional inputs (e.g., labor force and gross capita) and environmental variables, respectively. According to previous studies (e.g., O'Donnell 2018), TFP change is decomposed into several components, as follows:

$$TFP{I}_{lsit}=\underset{OTI}{\underbrace{\frac{\text{exp}\left(\lambda t\right)}{\text{exp}\left(\lambda s\right)}}}\underset{OEI}{\underbrace{\frac{\text{exp}\left({\phi }_i\right)}{\text{exp}\left({\phi }_k\right)}{\prod }_{j=1}^{J^{*}}{\left(\frac{z_{jit}}{z_{jks}}\right)}^{{\delta }_j}}}\underset{OSEI}{\underbrace{{\prod }_{m=1}^M{\left(\frac{x_{mit}}{x_{mks}}\right)}^{{}^{{\beta }_m\left(\frac{r-1}{r}\right)}}}}\underset{OTE}{\underbrace{\frac{\text{exp}\left(-u_{it}\right)}{\text{exp}\left(-u_{ks}\right)}}}\underset{SNI}{\underbrace{\frac{\text{exp}\left(v_{it}\right)}{\text{exp}\left(v_{ks}\right)}}}$$ (5)

To figure out how government spending affects TFP change and its components, we use the following equation, guided by the literature (e.g., Su and Nguyen 2022, Thanh et al. 2020).

$$Y_{it}={\beta }_0+{\beta }_1.GS{P}_{1it}+{\beta }_2.GS{P}_{2it}+{\beta }_3.GS{P}_{3it}+\sum _{k=4}^K{\beta }_{k}Contro{l}_{ikt}+{\delta }_i+\lambda t+{\epsilon }_{it}$$ (6)

where, Y_it indicates the TFPI index and its components, in country i in period t. GSP_it reflects government spending on health, education, and the military. Since government spending can be endogenous, instrumental variable approaches are used for estimations. In this study, according to previous studies (e.g., Fisman and Svensson 2007, Van Huong and Cuong, 2019), the mean values of government spending according to region, sector, and year are used as instrumental variables. Control_ikt is a vector of control variables selected in accordance with previous studies (e.g., Su and Nguyen 2022), trade openness, natural endowment, digital index (e.g., mobile connectivity), and institutional quality variables. δ_i denotes invariant unobservable variables and ε_it is the error term.

4. Empirical results and discussion

Using instrumental variable estimations or the two-system GMM estimator, Table 1 shows how different types of government spending affect TFPI. The results show that spending money on health impacts significantly TFP, because there is a high risk of pandemics in these countries, their economies are not very strong, and they have large populations and low standards of living (Bhorat et al., 2021; Kim and Ahn, 2020). Deaton (2013) says that health, which is a reflection of human ability, is the basis for human progress and development. As a result, investing in health helps improve the well-being of citizens, reduces inequality, and makes workers more productive.

Table 1.

The impact of government spending on productivity and its decompositions.

Variables	TFPI			OEI
	(1)	(2)	(3)	(4)	(5)	(6)
Government spending on health	0.387***			0.010
	(0.138)			(0.018)
Government spending on the military		−0.024**			0.0001
		(0.011)			(0.003)
Government spending on education			0.003			−0.007**
			(0.012)			(0.003)
Trade size	−0.054*	−0.015	0.035	−0.001	−0.006	−0.012
	(0.032)	(0.037)	(0.025)	(0.003)	(0.008)	(0.008)
Natural resources	−1.012**	0.023***	0.014*	0.186	0.001	0.006***
	(0.453)	(0.009)	(0.008)	(0.157)	(0.002)	(0.002)
Mobile Connectivity Index	0.004***	0.003***	0.000	0.000	0.000	0.001***
	(0.001)	(0.001)	(0.001)	(0.000)	(0.000)	(0.000)
Voice and accountability	−0.021	−0.021	−0.019	0.001	−0.004	−0.013***
	(0.033)	(0.023)	(0.015)	(0.003)	(0.005)	(0.005)
Political stability and absence of violence/terrorism	0.019	−0.066*	−0.010	0.008***	0.014**	0.007
	(0.017)	(0.037)	(0.020)	(0.002)	(0.006)	(0.005)
Rule of law	−0.028	0.031	−0.037	0.084***	0.006	0.009
	(0.025)	(0.056)	(0.031)	(0.004)	(0.011)	(0.010)
Control of corruption	0.074***	0.007	0.053*	0.019***	−0.018	−0.009
	(0.025)	(0.060)	(0.028)	(0.004)	(0.012)	(0.008)
Constant		1.228***	0.586**		0.997***	1.128***
		(0.325)	(0.271)		(0.077)	(0.075)
Observations	887	926	706	887	926	706
Number of panels	149	137	139	149	137	139
Weak identification test and [Stock-Yogo weak ID test critical value at 10%]	28.305 [16.38]			28.305 [16.38]
Hansen test of over-identification (p-value)		0.316	0.343		0.385	0.339
Diff-in-Hansen tests of exogeneity		0.301	0.275		0.374	0.341
Notes: Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. Models also control for year dummies.

In most economic clusters, TFP goes down when money is spent on the military. There is a trade-off (guns or butter) between spending on the military and spending on other things. If there are no threats to territorial integrity, spending on the military will lower TFP, especially in transitional countries (Feng et al., 2017; Zhao et al., 2017; Su and Nguyen, 2022). There are at least two other ways that military spending can hurt the economy: (i) crowding out effects and (ii) sending misleading market signals.

The way government spending affects the decompositions of TFP change (like OSEI and OEI) can be ascertained from Table 2 . First, military spending hurts TFPI mainly because it makes scale and allocation efficiency (OSEI) worse (columns 3 and 6 in Table 2). This insight confirms what has already been discussed. The problem is especially severe in countries undergoing change, like Ukraine, where war and lack of funds have destroyed the structure of many manufacturing industries. Spending on health care is good for the SNI in transitional countries (Table 2, column 4). Extra spending can also make it harder for private companies to do business and sends misleading economic signals. It is also important to remember that the healthcare system can only change when things are normal. When COVID-19 and other shocks occur, the system becomes overloaded and this harms the economy (Armocida et al., 2020). McKibbin and Fernando (2021) say that the damage caused by COVID could have been lessened if the health system had more funding when the peak of the epidemic was reached.

Table 2.

The impact of government spending on productivity's decompositions.

Variables	OSEI			SNI
	(1)	(2)	(3)	(4)	(5)	(6)
Government spending on health	0.002			0.354***
	(0.043)			(0.101)
Government spending on education		−0.004			0.016
		(0.005)			(0.010)
Government spending on the military			−0.008**			−0.016*
			(0.004)			(0.010)
Trade size	−0.006	0.018*	−0.003	−0.046*	0.032	0.001
	(0.008)	(0.010)	(0.015)	(0.025)	(0.024)	(0.028)
Natural resource	−0.405**	0.004	0.007**	−0.726*	0.003	0.017*
	(0.162)	(0.003)	(0.003)	(0.414)	(0.006)	(0.009)
Mobile Connectivity Index	−0.000	0.001	0.001***	0.004***	−0.001	0.002
	(0.000)	(0.000)	(0.000)	(0.001)	(0.001)	(0.001)
Voice and accountability	−0.003	0.006	−0.008	−0.019	−0.011	−0.010
	(0.009)	(0.006)	(0.007)	(0.025)	(0.014)	(0.019)
Political stability and absence of violence/terrorism	−0.003	−0.009	−0.016*	0.014	−0.006	−0.057**
	(0.005)	(0.006)	(0.010)	(0.013)	(0.016)	(0.024)
Rule of law	−0.013*	−0.031***	−0.012	−0.098***	−0.021	0.039
	(0.008)	(0.010)	(0.016)	(0.019)	(0.030)	(0.049)
Control of corruption	0.016*	0.028***	0.020	0.034*	0.035	0.005
	(0.008)	(0.010)	(0.014)	(0.021)	(0.028)	(0.051)
Constant		0.950***	1.071***		0.461*	1.113***
		(0.115)	(0.134)		(0.254)	(0.265)
Observations	887	706	926	887	706	926
Number of panels	149	139	137	149	139	137
Weak identification test and [Stock-Yogo weak ID test critical value at 10%]	28.305 [16.38]			28.305 [16.38]
Hansen test of over-identification (p-value)		0.429	0.340		0.452	0.366
Diff-in-Hansen tests of exogeneity		0.619	0.1		0.454	0.1
Notes: Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. Models also control for year dummies.

5. Summary and policy implications

This study has examined the way government spending affects TFP and its decompositions, "momentum" for the growth of the "Asian Miracle," and "cushioning" in times of crisis. World productivity has gone up, but the productivity growth rate has gone down substantially, especially since the COVID-19 pandemic. Allocative efficiency and scale were very important to improving productivity before 2019. Increased spending on the military hurts both the allocation efficiency and environmental efficiency of a country's productivity. On the other hand, spending on health improves a country's productivity, especially in transitional countries.

Our research shows that countries must improve their economy's allocation of resources through international trade and give the private sector greater, clear access to resources. Furthermore, if transitional countries seek to improve their TFP, their governments must put more money into the health sector. Transitional countries often have weak health systems, low living standards, and a large population, which make them vulnerable to further pandemics. Moreover, spending on the military is expensive and sends misleading economic signals. Limiting conflicts and military escalation will be important for future economies to grow in a sustainable way.

Declaration of Competing Interest

The authors agree that this research was conducted in the absence of any self-benefits, commercial or financial conflicts and declare absence of conflicting interests with the funders.

Appendix 1. TFPI and changes in its components

Year	TFPI	OTI/TECH	OEI	OSEI	OTEI	GEI	RQI	SNI
2014	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
2015	1.006	1.013	0.999	0.993	1.000	0.999	1.000	1.001
2016	1.019	1.027	0.998	0.987	1.000	0.999	1.000	1.008
2017	1.036	1.040	0.998	0.980	1.000	0.999	1.000	1.018
2018	1.047	1.054	0.998	0.973	1.000	0.998	1.000	1.023
2019	1.057	1.068	0.998	0.967	1.000	0.998	1.000	1.026
2020	1.034	1.082	0.998	0.976	1.000	0.998	0.999	0.981
	dTFPI	dTECH	dOEI	dOSEI	dOTEI	dGEI	dRQI	dSNI
15/14	1.006	1.013	0.999	0.993	1.000	0.999	1.000	1.001
16/15	1.013	1.013	1.000	0.994	1.000	1.000	1.000	1.006
17/16	1.016	1.013	1.000	0.994	1.000	1.000	1.000	1.010
18/17	1.011	1.013	1.000	0.992	1.000	1.000	1.000	1.006
19/18	1.009	1.013	1.000	0.994	1.000	1.000	1.000	1.002
20/19	0.978	1.013	0.999	1.010	1.000	1.000	0.999	0.957
20/14	1.034	1.082	0.998	0.976	1.000	0.998	0.999	0.981

Source: Authors.

Data availability

• The authors do not have permission to share data.