An Assessment of the Nexus Between Government Expenditure and Inflation in Nigeria

Abstract Research Background: The on-going debate concerning the exact relationship that exists between inflation and government expenditure especially in the long and short run prompted this research. Purpose: The study assesses the relationship between government expenditure and inflation in Nigeria. Apart from government expenditure and inflation rate, other variables such as exchange rate and money supply are included to ensure a robust model. Research Methodology: Secondary data from 1980 to 2017 were collected and analysed using the Johansen Cointegration analysis and vector error correction model. Results: The results showed that apart from the bi-directional relationship that exists between the variables, there exists a strong relationship between government expenditure and inflation rate and that a significant impact is sustained from the short run through the long run. The exchange rate and money supply also exhibit a strong association with government expenditure. Novelty: The study has underscored the importance of the inflation rate in Nigeria as it affects government spending by focusing more on inflation rather than the movement that was the focus of most of the previous studies. It has also shown the causality flow from both inflation and government expenditure, which hitherto remains contentious.


Introduction
In both developed and developing countries, there is concern for raising the standard of living over time, but this need is more pronounced in developing countries given the extent and depth of poverty in these countries. The implication of this is that government at all levels needs to provide incentives for investment via government spending. In Nigeria, government spending rose by about 39.5% in 2009, this value was almost doubled between 2013 and 2014 at the peak of the increase in oil price (CBN, 2015). The implication of increase in government spending is directly felt on the money supply (Abu, Abdullahi, 2010).
At the twilight of bank recapitalization in Nigeria between 2006 and 2008, there was an increase in money supply when the government increased spending to increase the deposit base of some commercial banks and consequently increase money supply (NDIC, 2011).
Theoretically, the expected overall effect is the general upward price movement of goods and services in the economy (often caused by an increase in the supply of money), usually as measured by the consumers' price index and the producer price index. Overtime, as the cost of goods and services increases, the value of the currency falls because a person cannot purchase as much with that amount as he or she previously could do. This shows that there is a likelihood of inflation when money supply rises.
Considering the trend of inflation in Nigeria, it appears that there were periods when the Naira was very strong in 1970 and yet there was a high inflation rate then due to an increase in money supply as a result of the rise in oil price. However, the situation appears to be different now because since the inception of the fourth republic till 2010, the situation of inflation in Nigeria has assumed epidemic proportions due to excessive government spending (Olayemi, 2010). This scenario has thrown up a debate as to the correct relationship that exists between inflation and government spending.
These debates rest mainly on economic theories and some empirical findings. The first line of thought is the aggregate demand and supply theories as propounded by the Keynesian school. From the theory, when aggregate demand rises as a result of an increase in consumption and investment in the private and public sector occasioned by a rise in government expenditure, there will be upward pressure on the general price level. On the other way round, inflation can affect government expenditure and even cause it to be negative and then it is viewed from the supply side where a government issues bonds. This will put pressure on interest rate and hence cause investment and output to fall. At this point any expansionary government spending will not have the desired effect and it may even have a negative multiplier effect as a result of the rise in the general price level especially if the crowing out effect is large (Romer, 1996). These two views have been supported by the following empirical studies (Odusola, Akinlo, 2001;Oniore, Obumneke, Torbira, 2015;Tai, 2014).
Based on the discussions above, it appears that there are two schools of thought. Some believe that increase in government spending may lead to a rise in inflation but other authors believe that the situation can only exist in the short run because the influence of government expenditure on boosting output will in the long run lead to a fall in inflation (Wooldridge, 2013;Williams, Adedeji, 2004). However, the debate on inflation growth and government expenditure nexus is still on going. The argument had centered on whether or not increasing public spending has the potential to induce inflation (Ezirim et al, 2006). Consequently, there is still an unresolved issue theoretically as well as empirically as to the effect of government spending on inflation.
Again, the direction of causality between the two has also generated a lot of debates. While V. Piana (2001) agreed that government spending has a bidirectional relationship with inflation, J.O. Oniore, E. Obumneke and M.T. Torbira (2015) are of the contrary opinion that it is just a unidirectional causality that flows from government expenditure to inflation.
Considering the explained lack of consensus regarding the relationship between inflation and government spending, this study will be contributing to the existing literature by investigating this using Nigerian historical data from 1980 to 2017. The remaining part of the paper is divided into methodology, results and discussion, and finally conclusions.

Literature review
Although there are many empirical literatures on inflation and government expenditure relationships the few that are on Nigeria are discussed as follows.

D.O. Olayungbo (2013) examined an asymmetry causal relationship between government
spending and inflation in Nigeria from the period of 1970 to 2010. The asymmetry causality test shows that a uni-directional causality exists from negative government expenditure changes (low or contractionary government spending) to positive inflation changes (high inflation) in the Vector Autoregression (VAR) model. The finding implies that inflationary pressure in Nigeria is state dependent, that is high inflation is caused by low or contractionary government spending. It is evident from the reviewed literatures that there are lack of consensus on what exactly is the relationship between inflation and government expenditure in Nigeria. This study apart from contributing to the growing literatures on this will also focus more on inflation rather than the movement that was the focus of most of the papers reviewed above.

Model specification
Leveraging on the aggregate demand and supply theory of Keynes, which has been tested and applied in different studies such as A.C. Pigou (1989), this study will include money supply and exchange rate as additional control variables. Therefore, m odel specification in the study consists of a system of o n e equation, which includes three explanatory variables. The model is specified thus: where: GEXP t -government spending at period t (proxies as total expenditure as a percentage of GDP), EXR t -exchange rate at period t (proxies as real exchange rate), MS 2t -broad money supply at period t (proxies as broad money supply as a percentage of GDP), -stochastic variable (error term), -constant term, θ, β and γ -parameters to be estimated.

Method of analysis
Quantitative method is applied to analyse the data. The time series analysis is embraced and it involves the following procedures: unit root or stationarity tests, co-integration tests, and Granger Causality tests. The details of these are presented as follows:

Unit root test
This procedure is part of the pre-estimation test necessary for most time series estimations.
Since many macroeconomic data are largely non stationary therefore, there is the need to make them stationary so as to make them suitable for estimation and to guide against spurious regression. In this study the Augmented Dickey-Fuller (ADF) Techniques are used to test and verify the unit root property of the series and stationarity of the model (Dickey and Fuller, 1997).

Co-integration test
It is possible for a non-stationary variable to have a long run stationary relationship hence the application of cointegration (Gujarati, 2004, p. 167;Yang, 2000, p. 78). The cointegration test is important to examine the existence of co-movement among or between two variables, in this case inflation and government expenditure. The study uses the Johansen co-integration test to for long run relationship since it is the one of the prominent tests which can estimate more than tone co-integration relationship if the data set contains two or more time series as well as gives the maximum rank of co-integration (Johansen, Juselius, 1994).

Causality test
In order to determine which variable in the model causes the other, the Granger causality test is used. The F-statistics are used to reject or accept the null hypothesis of no causation between the variables. The granger specification is described in equations 2 and 3.
The reverse is stated thus: where: GEXP is government expenditure, INF is inflation rate, m's are lag periods, β 1j and β 2j are parameters to be estimated and µ t denotes the stochastic error term.

Error correction model (ECM)
The Engle -Granger representation theorem proves that, if a co-integrating relationship exists among a set of I(1) series, then a dynamic error-correction (EC) representation of the data also exists. The methodology used to find this representation follows the "general-tospecific" paradigm (Hendry, 1987). Initially, the first difference of each variable in the model for this study, a constant term, and a one-lagged EC term (EC t-1 ) generated from the static regression procedure were used. Then the dimensions of the parameter space were reduced to a final parsimonious specification by sequentially imposing statistically insignificant restrictions or eliminating insignificant coefficients (Williams, Adedeji, 2004 Where GEXP is government expenditure, INF is inflation rate, EXR is exchange rate, MS 2 is broad money supply, ECM is error correction term (i.e. measure of the speed of adjustment), Δ is the first-difference operator, p's are lag periods, β 0 is constant, β 1 -β 5 are parameters to be estimated and µ t denotes the stochastic error term.

Sources of data and measurement
The data used for this study are basically time series data covering 1980-2017, that is thirty-eight (38) years. The data were sourced from the Central Bank of Nigeria (CBN) Statistical Bulletins. In terms of measurement, government total expenditure as percentage of GDP is used to proxy government spending, inflation rate is used; real exchange rate is used for the exchange rate while broad money supply (MS 2 ) as a percentage of gross domestic product (GDP) is used as a proxy for money supply.

Results and discussions
This section involves the results and interpretation of the unit root test, co-integration test, Granger causality test and vector error correction model.

Unit root test
In order to estimate the vector error correction model, the variables must be free from unit root problems, meaning that they have to be stationary at the same order of integration.
The software for the ADF test automatically selects the optimal lag length of 2. Therefore, the result of the Augmented Dickey-Fuller Unit Root Test is presented as follows: Step model estimation approach that the variables do not have unit a root problem at the integration of order one, I(1). On the basis of this, a null hypothesis of being non-stationary is rejected and it is safe to conclude that the variables are stationary.

Co-integration test
The Johansen Co-integration test was carried out to determine the long run relationship among the variables in the model. The trace statistics and the maximum Eigen value are compared with the Mackinnon critical value at 5% of significance in order to determine the number of co integrating vector equations in the model using one lag interval with intercept and trend. The result of the Johansen co-integration Test is presented below in Table 2. From Table 2, the trace statistics exceeds the critical value at the 5% level of significance for hypotheses rank 0, meaning that the hypothesis that there is no co-integration equation or error term in the model is rejected. This is ascertained, as a critical value is less than the trace statistics at none rank of co-integration. These results showed that only one co-integrating equation exists among the variables. Therefore it is shown that there is an existence of a longrun dynamic relationship among the variables and an Error Correction Model is thus justified.

Co-integration equation
After the establishment of a long run relationship via the co-integration test, there is the need to estimate the long run equation that shows the relationship among the variables. The long run relationship is presented below in Table 3.
The relationship between government expenditure and inflation in the long run is described in Table 3. In addition, the results include the relationship in the long run between government expenditure and other variables such as exchange rate and money supply. The results give the coefficient of inflation in the long run to 0.038685. This value is significant at 5% meaning that inflation as a variable has a significant impact on government expenditure in the long run. Since the coefficient is positive, it indicates that a rise in the inflation rate will cause government expenditure to rise in the long run. The exchange rate also has the same relationship with government expenditure showing a coefficient of -0.070847. However, there is a significant inverse relationship between the two. It simply means that as the exchange rate is rising, government expenditure will be falling significantly. Money supply is also relatively significant in the model but at 10%. The implication is that money supply has a direct and significant relationship with government expenditure.

The Granger Causality test
The test is carried out using the Pairwise Granger Causality method and the maximal lag difference chosen is 1 in order to make the test effective. Again, the causality test as illustrated in equations 2 and 3 only involves the two core variables in the study namely; inflation rate and government expenditure.  It is shown from the results in Table 4 that the statements that inflation (INF) does not granger cause government spending (GEXP) and GEXP does not granger cause INF are both rejected at 5%; the F statistics of both have probabilities that are less than 0.05. Therefore, we can conclude that there exists a bidirectional causality between inflation and government expenditure.

Estimation of the error correction model
According to the Granger Representation theorem, when variables are co-integrated, there must also be an error correction model that describes the short-run dynamics or adjustments of the co-integrated variables towards their long-run equilibrium values. ECM consists of one-period lagged co-integrating equation and the lagged first differences of the endogenous variables. Using the Vector Auto-regression (VAR) method with a lag interval of 1 to 2 and intercept (no trend) in co-integration (CE), the ECM is estimated as can be seen below with government spending (GEXP) as a dependent variable and the rest of the variables are said to offer an explanation to government expenditure behaviour in Nigeria. The result of the error correction model is shown above in Table 4. The speed of adjustment coefficient of the ECM is -1.7128 and it is significant at 1%. This implies that at every interval that is whenever there is disequilibrium, the recovery of the government spending back to equilibrium is 71.28%.
The implication of this is that there is a short run dynamics running from the inflation rate, exchange rate and broad money supply to government spending in Nigeria.

Short-run causality relationship
The result from this study as is shown above in Table 5 shows that government spending is statistically significant for the lagged periods in the short-run dynamic associations. The estimation reveals that the lag values of government spending are significant. Again, inflation exhibits a significant short run relationship with government expenditure. This result further affirms the strong association between government expenditure and inflation in Nigeria. Exchange rate and money supply in the same vein show a strong short run association with government expenditure.
The implication of this result is that the impact of inflation of government expenditure on the inflation rate is sustained from the short run period through the long run period.

Model overall significance
This is deduced from the F-statistics probability. Table 5 shows that the F-statistics probability is 0.00, which indicates that all the explanatory variables, will significantly influence government spending. Therefore, inflation rate, exchange rate and money supply are important determinants of government expenditure in Nigeria.

Model goodness of fit
This is estimated by using the R 2 , which shows a value of 0.7064 in Table 5. This value indicates that 70.64% systemic variations in government expenditure is explained by the inflation rate, exchange rate and broad money supply. This also indicates that the remaining 29.36% not captured by the explanatory variables in the model is due to changes in other variables or error terms.

Autocorrelation test
Two methods are used for the autocorrelation test in the study. Firstly, the correlogram approach via ACF and PACF are applied and the Breusch-Godfrey Serial Correlation LM Test.
Their results are shown in the following tables.
From both tests shown in Table 6, it is obvious that the estimated model does not have an autocorrelation problem. All the probabilities of both the ACF and PACF are not significant.
In addition, the Breusch-Godfrey Serial Correlation test also has the probability of the F statistics to be greater than 0.05 thus, showing that the null hypothesis of no autocorrelation is accepted at a 5% significant level.

Conclusions
The results from the analysis further underscore the importance of the inflation rate in relationship shows that rising inflation causes a government to spend more. The implication of this conclusion is that an increase in inflation rate has the tendency of increasing the prices of goods and service on which government expenditure is expended. Hence, there is the need for a government to jack up its expenses in order to meet the initial quantity of commodity budgeted.
A.M.O. Anyafo (1996) opined that the incessant upward review of the budget allocated for various infrastructures in Nigeria is as a result of the persistent rise in the general price level.
It can also be concluded from the study that apart from the inflation rate, exchange rate also exerts a significant impact on government expenditure. Findings from the study reveal that currency depreciation that is a fall in the value of the naira will have the tendency of increasing government expenditure. The mechanism through which this happens is very obvious as explained by Y.K. Adamson (2000). According to him, more naira will have to be sacrificed for the same commodity denominated in foreign currency e.g. US dollars whenever the naira depreciates. The implication is that the government will have to increase its spending to cover up for the increase in the naira. Nigeria remains the largest importer in the whole of Sub Sahara Africa; therefore, any devaluation is bound to have a significant impact on government expenditure.
Money supply is also an important factor that affects government expenditure. An upward movement in money supply in Nigeria influences government expenditure positively. The study found a strong positive relationship between money supply and government expenditure.
The largest spender in Nigeria is the government. This is because the economy is not driven by the private sector but largely, by the government.
This study has further shown the causality flow from both inflation and government expenditure. The implication is that government expenditure exerts a great influence on inflation pressure in Nigeria and vice versa. It is recommended that government policy on its expenditure should target output increase so that it can cushion the effect of the resulting inflation trend.