An investigation into the effect of budget deficit on the economic performance of Zimbabwe

ABSTRACT

The study investigated the impact of budget deficit on the economic performance of Zimbabwe for the period 2000–16. Using secondary data sourced from the International Monetary Fund (IMF), Zimbabwe National Statistics Agents (ZIMSTATS), World Bank and the Reserve Bank of Zimbabwe (RBZ) websites, the study estimated the relationship between budget deficit and economic performance using ordinary least squares (OLS) methods. The estimated results revealed a negative and significant relationship between budget deficit and economic performance in Zimbabwe. The study inferred that Zimbabwe’s huge budget deficit could be driven by recurrent expenditure such as salaries. We recommend the need for the government to have fiscal discipline and to channel resources towards the productive sectors of the economy such as capital projects for long term development.

KEYWORDS:

• Budget deficit
• Zimbabwe
• ordinary least squares

1. Introduction

Budget deficit, which is an excess of government expenditure over revenue, is one of the key national economic indicators. The size and the financing of the budget deficit is of concern to economists, policy makers and investors. A healthy deficit for a country should not exceed the prudent limit of 3% of the gross domestic product (GDP). Consequently, a budget deficit of less than 3% is desirable. In some cases, the budget deficit may exceed the prudent 3% without any adverse effects to the national economy. This is possible in countries where the deficit is used to finance capital expenditure like the United State of America and China, among developed countries. Some economic performance and deficit rankings are shown in Table 1.

Table 1. Budget deficit and economic ranking for selected countries

Rank	Country	% of budget deficit to GDP
1	USA	14.9
2	China	13.6
3	Japan	12.2
4	Venezuela	58.2
5	United Kingdom	9.2
–	–	–
120	Zimbabwe	12.8

Source: World Bank 2017

From Table 1, Zimbabwe’s deficit is lower than some of the developed countries such the United States of America (USA) yet, the economy of Zimbabwe is ranked amongst the worst with an economic ranking of 120. Given the size of Zimbabwe’s deficit relative to other countries, one would expect the ranking to be much better. A worrying development has been the dependence of Zimbabwe on countries with a higher deficit than her, such as China, for infrastructural development and other capital programmes.

Zimbabwe had been experiencing a downward trajectory for over a decade and had one of the horrible cases of hyperinflation which resulted in the suspension of the local currency. On the other hand the budget deficit maintained an upward trend which signified worsening fiscal performance. For countries like the USA, a widening budget deficit is not associated with poor economic performance. This means that the economy can grow in the presence of a budget deficit. We can therefore infer that there is no universal relationship between economic performance and the size of the budget deficit.

Given the ambiguous relationship between economic growth and budget deficit, the study aimed at unveiling the direction of the relationship between economic growth and the budget deficit. The other objective of the study was to come up with deficit management policies that promote Zimbabwe’s economic performance.

There has been research on Zimbabwe’s economic performance as influenced by the components of GDP, which are consumption, investment, government expenditure and net exports. Zuze (2016), Makochekwana (2010) and Mandishekwa (2014) though studied Zimbabwe’s budget deficit, their focus was on the political aspects of the deficit. To the best of our knowledge literature on the impact of the budget deficit on economic growth in Zimbabwe is deaf and thus our research was focused on the budget deficit and economic relationship in Zimbabwe.

This section gave the introduction of this research, highlighted the problem statement and the objectives of the study. The succeeding sections of the study focus on the literature review, methodology, result presentation and analysis and policy recommendations.

1.1. Literature review

Budget deficit is one of the indicators of economic performance. Firstly, a huge budget deficit shows that the economy is performing badly in terms of revenue collection. Budget deficit may also indicate fiscal indiscipline. There are various theories that can be used to explain budget deficit and some of these theories are discussed below.

1.1.1. The standard approach to budget deficit

The standard approach to budget deficits assumes that the substitution of a budget deficit for current taxation results in an increase in aggregate demand by consumers (Barro, 1989). This means that if the level of savings increases by an amount less than the tax cut this leads to a decrease in national savings. It applies in an autarky where there would be a need for predicted real interest rate increase to restore equilibrium between favourable national savings and the demand for investments. The increase in the real interest rate reduces investments, which is seen in the long run as a smaller stock of productive capital. Therefore, public debt is an intergenerational burden in the sense that it leads to an insignificant stock of capital for upcoming generations.

According to Futagami & Shibata (2003) in an open economy, a small budget deficit or any increase in public expenditure would have minor effects on the real interest rates in the international capital markets. Therefore, the home country’s resolution to a budget deficit for current taxes leads principally to increased borrowings from other countries, rather than to higher real interest rates. This means that the budget deficit leads to a current-account deficit. In most cases, there is a tendency of the country’s budget deficit to drive out its domestic investments in the short run. This would result in a decline in economic activities. In the case of Zimbabwe, research has shown no direct link between interest rates and the budget deficit as economic growth has been found to be influenced by the political environment (Kanyenze et al. 2017). Being highly indebted coupled with a politically unfavourable environment, much would need to be done by Zimbabwe to attract foreign direct investment, which is believed to be key in deficit reduction through improved economic performance.

1.1.2. The Ricardian approach/Ricardian equivalence

According to Rohn (2010), the Ricardian equivalence theorem assumes that individuals are forward looking and rational regarding the fiscal affairs of the government. It says that for a given track of fiscal expenditure, a deficit- financed by current tax cuts results in higher future taxes. This outcome is the result of the government budget constraint which says that total expenditures for each period should be equal to revenues from taxation or other sources. Holding fixed the track record of government expenditures and non-tax revenues, a cut in current taxes must be followed by an equivalent increase in the present value of future taxes (Barro 1989).

According to de Mello et al. (2004), the Ricardian equivalence principle says that people view the issue of fiscal debt and current taxes as the same. It therefore rejects any wealth effects from the issue of debt, people would willingly hold bonds in order to pay the higher future taxes caused by the deficit.

The Ricardian equivalence theorem argues that budget deficit does not affect the aggregate demand in an economy. Even if the bonds which are issued by the government to the public in order to finance the budget deficit are viewed as a wealth asset by individuals, the income from those bonds does not increase demand as the income would be offset by the tax which would be paid to the government by the private investors for holding government bonds. Therefore the Ricardian equivalence concludes that the bond financed budget deficit does not have an effect on the economic performance as the effects offset each other. From this theorem, we can infer that the expectations of future tax increases might hold back economic expansion as economic agents are generally tax averse. This situation is worse in Zimbabwe where more than 60% of the economy is informal, (Zimbabwe National Statistics Agency, 2019).

1.2. The tax-smoothing model

According to Pinho et al. (2004) the tax-smoothing theory of the government budget acts as a normative standard from which political economy models are derived, although some researchers view this approach as a description of actual fiscal policy. In fact, most of the recent political models are positive clarifications of the observed deviations from the tax-smoothing approach (Pinho et al. 2004). In general, positive theories of budget deficits try to explain the alterations between normative predictions and empirical evidence by relaxing the assumptions of the normative theory concerning actor’s preferences or institutional arrangements.

The model made an assumption of a closed economy with less capital. Taxes are distortionary in the economy because they affect labour supply, individuals are not willing to supply more of their labour especially if the taxes are progressive. It can be concluded that where taxation is progressive, individuals have a habit of under declaring their taxable income in order to evade higher taxes that are associated with higher budget deficit thus resulting in higher incidences of evasion and avoidance which is a common scene in Zimbabwe. The 2020 Budget Statement.

On the other hand, a tax-smoothing policy where the government is trying to maintain the tax rates at the same level would lead to a budget deficit in the current period and a surplus in the future, the surplus in the future would compensate for today’s budget deficit.

1.2.1. The Keynesian approach

The Keynesıan view gives allowance for the probability that some of the economic resources are unemployed. From this assumption, we can conclude that Zimbabwe’s revenue base was severely affected by massive deindustrialisation and idle farms following the controversial land reform and indigenisation policies. The theory also assumes that there is an availability of a large number of subjective individuals who are faced with liquidity constraint. This second assumption shows that individuals have a high marginal propensity to consume. In this case the Keynesians analysed budget deficit at an individual level, therefore in the case of governments it means that governments increase their public expenditure when their revenues increase. Many traditional Keynesians argue that deficits need not drive out private investment (Wu et al. 2014). Some suggest that an increase in aggregate demand changes the profitability of private investments and results in higher levels of investment at any given interest rate. This means that deficits may actually stimulate aggregate saving and investment regardless of the fact that they result in an increase in interest rates.

1.2.2. Empirical literature review

There has been a far-reaching argument over the past decade on the impact of budget deficit on economic growth.

Kosimbei (2009), using both qualitative and quantitative data from 1963 to 2007, analysed the effect of budget deficit in macroeconomic performance in Kenya. He analysed gross domestic product (GDP) as a function of balance of payments, budget deficit, private investment, private consumption, treasury bill rates and money supply. Applying vector auto regression (VAR) method, his study concluded that the budget deficit had a significant negative impact on economic performance. Using the index of macroeconomic stability, Kosimbei (2009) concluded that budget deficits have a significant effect on macroeconomic performance which he attributed to the budgetary process.

Studying the USA data, Fatas & Mihov (2005) revealed that large fiscal deficits result in increases in broad money supply which leads to an increase in the price level. The effect was through weakened exchange rates. Increases in exchange rates may lead to destabilise other macroeconomic variables such as investment, thereby negatively impacting economic growth (Kim & Roubin 2007).

Zuze (2016), using the VAR model on Zimbabwe data, highlighted that the presence of a negative impact of the budget deficit on economic growth since independence.

Mandishekwa et al. (2014), using Johansen co-integration and Granger causality, revealed that the twin deficits hypothesis holds in Zimbabwe. Makochekanwa (2010) highlighted the presence of a stable long run relationship between budget deficit, exchange rate, GDP and inflation.

This study is different from the reviewed studies in that they focused on the relationship of budget deficit to other macro-economic elements other than GDP. For example Makochekwana’s (2010) study focused on the relationship between budget deficit and inflation in Zimbabwe. The study by Mandishekwa et al. (2014) was on the twin deficits hypothesis. We looked at the relationship between budget deficit and GDP growth in Zimbabwe. Though this study is closely related to Zuze’s (2016) work, it differs in terms of the scope and the research methodologies which we used. Zuze used VAR and applied the ordinary least squares (OLS) methods.

2. Methodology

This section elaborated on the techniques used in data processing and analysis. The data was collected from secondary sources, in particular: International Monetary Fund (IMF) (2017), Zimbabwe National Statistics Agents (ZIMSTATS) (2016), World Bank (2017) and the Reserve Bank of Zimbabwe (RBZ) (2017) websites. A number of diagnostic tests were carried out to come up with results for analysis and drawing inferences.

Data analysis was done using the Ordinary Least Squares (OLS) methodology. The least squares estimation techniques choose the parameters which minimise the sum of the squared error terms. Before the final model was run, a series of diagnostic tests were done to validate the predictor variables.

The research estimated a linear model of economic performance as shown below $GDP={\text{β}}_0+{\text{β}}_{1}BD+{\text{β}}_{2}I+{\text{β}}_{3}C+{\text{β}}_{4}NX+{\mu }_t$Where:

• GDP – Economic growth

• BD – Budget Deficit

• I – Investments

• C – Consumption

• NX – Net Exports

• µ_t – Error Term

• µ_t is the random disturbance term.

• β₁, β₂, β₃ and β₄ are the coefficients which measure general influence of each predictor variable on GDP as a measure of economic performance.

2.1. Estimation procedure

Stata 10 software was used to estimate the model. Before the final model was estimated using OLS, a series of diagnostic tests were run.

2.2. Testing for stationarity

Testing for stationarity properties of time series is a very important exercise because use of non-stationary time series data in Classical Linear Regression Models (CLRM) as non-stationary data often leads to spurious results (Gujarati 2003). Unit root test was carried out using the Augmented Dickey-Fuller (ADF) test in order to evaluate the order of integration of each time series. The ADF test was used because it allowed us to process higher order data, since we used lagged variables. We also applied ADF because the series that we used was complex since our data was stationary at different levels. The variables that were not stationary in levels were differenced to make them stationary hence the Johansen co-integration test was carried out.

2.3. Lag length determination

This lag length was selected using the Akaike Information Criteria or Schwartz Information Criteria (Serena & Perron 2012). The lag length was essential for the implementation of unit root tests which was necessitated by the selection of an autoregressive truncation lag, say, n which was required in auto regression which we used to form the Dickey-Fuller and or the Durbin Watson test. We also required it in constructing an autoregressive estimate of the density at frequency zero.

2.4. Johansen co-integration test

This test was used to identify the number of co-integrating equations. The Johansen co-integration test was carried out to investigate the properties prior to the construction of an econometric model (Spider Financial Corp 2013). It acted as a descriptive tool to verify the stationarity of the time series data that we used.

2.5. Testing for multicollinearity

The data was tested for multicollinearity to correct for serial correlation between the variables. The presence of multicollinearity causes the model to be indeterminate.

2.6. Autocorrelation

Autocorrelation was tested using the Breusch–Godfrey (BG) test. It is the best test for autocorrelation because it allows for higher order autoregressive schemes and it also allows for higher order moving averages of the error terms.

3. Result presentation, analysis and interpretation

The study covered the period from 2000 to 2016. The period 2000 to 2008 data was manipulated using the annual Zimbabwean dollar to United States dollar exchange rates. This is because post 2008 Zimbabwe was using the United States dollar as the base currency.

3.1. Diagnostic tests

The results for the various diagnostic tests performed are shown below.

3.2. Johansen co-integration test

The Co-integration test was done to find out whether the variables are stationary at their level. The results are shown in Table 2.

Table 2. Co-integration results

Maximum Rank	Parms	LL	Eigenvalue	Trace Statistic	5% critical value
0	5	−672.18		86.2780	68.52
1	14	−655.24	0.8797	52.3913	47.21
2	21	−642.92	0.7855	27.7580*	29.68
3	26	−632.71	0.7209	7.3355	15.41
4	29	−629.87	0.2985	1.6638	3.76
5	30	−629.04	0.0987

Source: Own calculations.

From Table 2 at the maximum rank of 2 we concluded that there are at least 2 co-integrating equations. Because of the presence of co-integration, we went on to determine the lag length.

3.3. Lag length selection

Using the Akaike information criteria (AIC), the Hanan and Quinn information Criteria (HQIC) and the Schwartz-Bayesian information criteria (SBIC) the lag length was established and the results for each approach are as shown in Table 3.

Table 3. Lag selection

Lag	LL	LR	Df	p	FPE	AIC	HQIC	SBIC
0	−128.608				3.3e+06	17.8144	17.8118	18.0504
1	−118.57	20.074*	1	0.000	1.0e+06*	16.6094*	16.6064*	16.8926*
2	−118.568	0.00473	1	0.945	1.2e+06	16.7424	16.7389	17.0728
0	−110.87				3.4e+06	17.8262	17.7815	18.0435
1	−103.127	15.486*	1	0.000	1.2e+06	16.7888*	16.7352*	17.0495*
2	−103.126	0.00151	1	0.969	1.5e+06	16.9425	16.8800	17.2467
3	−102.922	0.40865	1	0.523	1.9e+06	17.0649	16.9934	17.4126
4	−102.63	0.58454	1	0.445	2.3e+06	17.1738	17.0934	17.5649

Source: Own calculations.

The results in Table 3 show the lag length selection which was tested with maximum lag order of 2 and at 99% confidence interval (1% level of significance). The results show that the optimal lag length is lag 1 as shown by the agreement of the AIC, HQIC and SBIC indicated by a *.

3.4. Testing for stationarity (unit root test)

Unit root test or test for stationarity was done using the Augmented Dickey-Fuller unit root test. Time series data is normally characterised by trends and is therefore in most cases non stationary. Therefore all the variables were tested for stationarity and those who were not stationary were differenced up to the first and second order differences. The first order differenced variable is shown by D1 and the second order is D2 as shown in Table 4.

Table 4. Stationarity test results

Variable	Lag	T-statistic	1% LOS	5% LOS	10% LOS	P- value
RGDP	0	0.504	−3.750	−3.000	−2.630	0.9850
RGDP	1	−0.255	−3.750	−3.000	−2.630	0.9317
RGDPD1	0	−2.410	−3.750	−3.000	−2.630	0.1389
RGDPD2	0	−5.527	−3.750	−3.000	−2.630	0.000*
Budget deficit	0	−2.649	−3.750	−3.000	−2.630	0.0834*
Investments	0	−3.237	−3.750	−3.000	−2.630	0.0179*
Net Exports	0	−1.296	−3.750	−3.000	−2.630	0.6310
Net ExportsD1	0	−4.488	−3.750	−3.000	−2.630	0.0002*
Consumption	0	−1.524	−3.750	−3.000	−2.630	0.5215
ConsumptionD1	0	−9.206	−3.750	−3.000	−2.630	0.0000*

Notes: *Indicates stationary data.

Source: Own Calculations.

3.5. Testing for multicollinearity

The variables were tested for multicollinearity and the results are as in Table 5.

Table 5. Multicollinearity results

Source	SS	Df	MS
Model	301082273	4	75270568.2
Residual	30563572.4	12	246964.37
Total	331645845	16	20727865.3

Notes: Number of observations = 17; F (4, 12) = 29.55; Prob > F = 0.0000; R- Squared = 0.9078; Adjusted R-Squared = 0.8771; Root MSE = 1595.9.

Source: Own Calculations.

There was no linear relationship between the explanatory variables.

3.6. Autocorrelation

Autocorrelation was tested using the Breusch–Godfrey LM test and the results were as in Table 6.

Table 6. Autocorrelation results

Lags (p)	Chi2	df	Prob > Chi2
1	0.178	1	0.6735

Source: Own Calculations.

In this test the chi-square value is 0.178 and the P value (Prob > chi2) is 0.6735 and we concluded that there was no serial correlation in the time series data in the study.

3.7. Result analysis

After the diagnostic tests were done and the necessary data manipulations to conform with the expectations of the Ordinary Least Squares regression, the OLS regression was run and the results output is presented in Table 7.

Table 7. Regression Results

RGDPD2	Coef	Std Err	t	P > | t |	95% Conf	Interval
Budget deficit	−2.59285	0.9905488	−2.62	0.022	−4.75107	−0.4346296
Consumption D1	0.728294	0.2146282	3.39	0.005	0.2606592	1.195929
Constant	869.8099	1109.801	0.78	0.448	−1548.239	3287.859

Source: Own Calculations.

In the regression results in Table 7 there were only two significant variables, which were budget deficit and consumption. This means that budget deficit and consumption determine economic growth (real GDP). This might imply that the deficit is financing recurrent instead of capital expenditure. A unit increase in consumption results in 0.7283 increase in real GDP which means that consumption has a positive effect on economic growth in Zimbabwe. This is consistent with the GDP and consumption relationship theories. These results are also consistent with the result by Odhiambo (2008), when using Tanzanian data.

The study revealed that budget deficit affects economic growth negatively. This is indicated by the negative coefficient of −2.592846. This means that a unit increase in budget deficit results in a 2.592846 decrease in real GDP. The results concurred with Kosimbei’s (2009) and Zuze’s (2016) findings who carried out their studies using the Granger causality and vector auto regressive estimation techniques respectively.

The results of this study also support the neoclassical theory which states that budget deficit is detrimental to an economy (Saleh 2003). The results support the neoclassical economist’s ideology which views budget deficits as having a negative effect on economic performance.

Investment and net exports were found to be insignificant in the model and thus were dropped.

4. Conclusion and recommendations

The research was interested in finding out whether budget deficit affects economic growth in Zimbabwe. The problem was ambiguity of the impact of budget deficit on economic performance. The main objective of the study was to assess the impact of budget deficit on economic growth.

Using the ordinary least squares methods and secondary data, the research unearthed the existence of a statistically significant negative impact of budget deficit on economic performance in Zimbabwe. Therefore, the regression results failed to reject the null hypothesis and it was concluded that budget deficit hinders economic growth in Zimbabwe. This could be because, in Zimbabwe, the deficit is channelled towards recurrent expenditure such as salaries, wages and social grants.

In light of the research findings, Zimbabwe’s fiscal policy has been characterised by large budget deficits which has had a negative effect on economic performance. Some policy prescriptions were recommended to minimise the impact of the budget deficit on economic performance.

The government should try to finance its expenditure through its revenues, thus fiscal discipline. The government revenue sources should be expanded by introducing more innovative ways of financing the budget. For example, infrastructure could be funded in partnership with the private sector under public private partnerships (PPPs) on a build operate and transfer (BOT) model.

The policymakers and parliament should play their roles in the budget setting process to ensure that individual interests do not outweigh national interests. The treasury should be apolitical and focus on budget items that foster economic progress. It should also consider the solid budget constraint of the government to achieve optimality.

The government should set a balanced budget and leave no allowance for deficit. The budgeting team could consider adopting a detailed performance based budget for its annual and subsequent budgets. This would help the authorities to closely monitor spending by government departments.

Future studies need to examine the sources of budget deficit in Zimbabwe so as to pluck out unnecessary expenditure.

Acknowledgement

We would like to acknowledge Lupane State University for according us the time to carry out this research. We would also like to acknowledge the input from our colleagues and all those who helped us with the much needed data.

Disclosure statement

No potential conflict of interest was reported by the author(s).