The Twin Deficit Hypothesis: An Empirical Analysis for Uganda

This study explored the validity of the twin deficit hypothesis in Uganda for the period 1980-2017. The methodology involved employing the Johansen co-integration test to find out whether there exists a long run link among variables current account deficit, fiscal deficit, RGDP, lending Interest rate and real effective exchange rate. The VECM is used to check how stable is the long run link between the variables while the granger causality test was done to conclude the direction of causality between current account deficit and fiscal deficit. The result confirms a long run link between all variables and we supported the Twin Divergence Hypothesis (TDivH) in Uganda since, VECM empirical results propose that fiscal deficit is negatively linked with current account deficit both in short run and long run and this is statistically significant at 5% level with a reasonably high speed of adjustment towards equilibrium. Increase in fiscal deficits improve current account deficits however  granger causality test result show that causality is reversed running from current account deficit to the budget deficit. Policy initiatives should be directed towards improving current account deficit through value addition to Agriculture products exported in order to increase foreign exchange earnings. Additionally, maintaining a conducive Marco-economic environment is essential, strengthening external policies, expanding employment generating projects, improving domestic infrastructure inform of roads, electricity supply, and proper storage facilities for agriculture products will further improve domestic production and minimize on massive importation of consumer goods that can be locally produced. If these policies are well implemented, current account deficit balance will improve. Lastly the government of Uganda needs to minimize on excessive non-development expenditures especially in political administrations that in most cases over use tax payers money with no or less profit. 

 

 

1. Introduction.

The major issue behind the twin deficit hypothesis is that fiscal expansion increases budget deficit as the government is trying to achieve its economic growth goals (Barro, 1989); this is accompanied by increase in people’s disposable incomes since taxes reduce making consumption to increase and lower national savings causing current account deficits. The lower savings cause government borrowing to further finance its expenditures. Furthermore, the twin deficit phenomena asserts that negative budget balances in a non-closed economy increases domestic interest rate and appreciates domestic currency which in turn depreciates current account balance. The impact of these budget deficits on current account balance, interest rates and exchange rate has attracted many researchers’ attention. Many researchers have found a long-run positive link between the budget and trade deficits using diversified econometric methods and samples while others find negative link between the two deficits (see Epaphra, 2017, Sakyi, 2016, Lau et al., 2010). The Ricardian Equivalence Hypothesis (REH) is one of the theoretical works that explains the link between negative budgetbalance and current account deficits. It postulates that the two deficits are not related at all, real interest rates are not affected by changes in taxes or budget deficits and neither does fiscal expansion affect current consumption and investment. On a contrary, the MundellFlemming model based on Keynesian analysis asserts that the two deficits are positively related with fiscal deficits causing current account deficits hence the twin deficit hypothesis.

These deficits lead to less foreign and government investment as investors don’t want to risk in the unstable economy and governments also tend to focus on debt repayments rather than investment. Future savings and consumption levels are also affected by the government debt. These deficits can shift from one individual country and affect the global financial system if they are un resolved hence commonly termed as the Macroeconomic problems of all economies(Mendoza et al., 2007).

Despite the current growth performance of Uganda, the country is facing internal limitation like negative fiscal balances andtrade deficits for the last two decades (Wambui, 2016). After the liberalization of trade in the mid 1980’s, the country involved in the trade of agriculture products across the world however the dependence on a limited variety of exports mainly coffee, tea, horticulture products that are semi-processed has made the country’s exports vulnerable to trade shocks hence persistent current account deficits to date (see Figure 1). Due to trade openness, the deficit was increasing specially in the years 2008, 2010 and 2011 that recorded the highest current account deficits while 2014-2017 current account balance is improving due to reduction in trade openness[1].The Ugandan shilling has also lost value in relation to other trading currencies (BOU, 2018). In addition, Loans obtained from abroad appreciate the exchange rates and affect the exporting sector hence current account deficits(Lwanga& Mawejje, 2014)

The increasing fiscal deficit for the past decade in Uganda is caused by low levels of revenue accompanied with increased government expenditure of over 20% of GDP. Government expenditure in the 1990s and early 2000’s was increasing greatly mainly due to increase in public wages and salaries in government workers, decentralization policies which involved introduction of new districts, civil servants among others all increased the government administrative costs during that period (Lwanga, & Mawejje, 2014). The National Development Plan (NDP1) states that between the years 2010&2015, infrastructure development within the countrywas the main cause of the fiscal deficit and it’s expected to reach 9% of GDP in 2019/2020.The country’s decline in growth rate from 2010 was to be boasted by embarking on massive infrastructure development (IMF, 2017). Furthermore, increased expenditures on military, energy and social programs that improve people’s quality of lifehave contributed to the negative budget balance. The government increased its allocations in the transport sector from $1.3 billion in 2016/17to $7 billion in 2017/18 while the energy sector allocations increased by 26.1 percent in 2017/18[2].

As most researchers have found out that in a non-closed economy budget deficits worsenscurrent account balance as postulated by Keynesian and therefore it’s important to find out whether there is a short or long run co-integration between Uganda’s fiscal and current account balancesas it helps the government to make adjustments in its fiscal expenditures in order to maintain a reasonable current account balance which in turn helps to draft proper macroeconomic policies.

In this context, the question worth investigating is whether there could be a long run association between a country’s fiscal and current account deficits. In order to attain a rightful conclusion and answer to this question, we investigated the (i) twin deficits hypothesis (increase in fiscal deficit worsens the external trade deficit), (ii) Ricardian equivalence (i.e. the two deficits are unrelated) or (iii) Twin divergence hypothesis (i.e. an increase in fiscal deficit improves current account deficit) hold. Furthermore, it’s of great importance to determine the direction of causality for better policy implementation.

Well some few researches have handled the case of Uganda and mixed results have been obtained for example, (Wambui, 2016) found a long run insignificant positive association between the two deficitswhile studying East African countries.(Ahmad &Aworinde, 2015), in their study about the TDH in sub-Saharan countries found a negative link between the deficits for Uganda yet(Mugume&Obwona, 1998) found no co-integration amongthe two deficits. On a contrary (Lwanga& Mawejje, 2014) found a unidirectional causal relationship running from budget deficits to current account deficits and raising interest rates hence confirming the Keynesian hypothesis.

Most of these studies above analyzed the co-integration between two variables Current account and fiscal balance. This study uses a model that comprises of five variables namely current account balance, fiscal balances with control variables namely real GDP, Lending interest rate and real effective exchange rate. All these variables are important in explaining the deficits whileReal GDP further shows the economic performance of the Ugandahence the study pursues to find whether there exists a short or long run association between variables during the period of study for future policymeasurements and suggestions.

Figure 1.Trend of fiscal and current account balance in Uganda as a percentage of GDP in the period(1980-2017)

 

Source:World Bank data.

Fiscal deficit has often led to capital outflow as most debts are paid in foreign currencies. In Uganda fiscal deficit increased from -2.6% of GDP in (1990-1999) to -3.8% of GDP in (2010-2017) Similarly, Current account deficit rose from -4.3% of GDP to -8.3% of GDP in the same time period(See Table 1).

The current account deficit was slightly lower and improving in the period of 2000-2009, it later increasedafter in 2010 mainly because of the developments in the exchange rates and reduced national savings. In 2014/15 financial year, the Ugandan Shilling was devalued 27% against the US dollar with trade weighted rate index depreciation of 17.5% accompanied by falling export item prices[3].The country is the net exporter of Coffee and any shortages or decline in its exportation affects the exchange rate (Mugume&Obwona, 1998). Furthermore, the growing private sector in the country is importing a lot of capital goods whose demand does not respond to price changeshence capital outflows. (Wambui, 2016) explains that government internal and external borrowing has worsened current account deficits.The gross national savings of Uganda reduced from 35.2% of GDP in 1980 to 21.6% in 2018 and this is estimated to increase to 23% in 2021 (IMF, 2018).

The fiscal surplus in 2003-2005 might be due to various debt relief programs[4] and the country experienced a current account surplus within the same period. Besides that, it has adopted policy reforms to have favorable Marco economic performance and also joinedthe East African Monetary union whose criterion is maintaining a less than 3% deficit. These reforms include mainly spending and revenue management reforms (bringing in new taxes like mobile money and social media tax in 2018 among others) in order to finance the public debt whichrose to 38.2% of GDP in 2017 from 23.5% in 2010 (IMF, world economic outlook 2018). These Fiscal policy reforms however affectthe behavior of private consumption and changes in real GDP of a country as consumers might cut the consumption levels because of reduction in their disposal income.

 

 

Table 1: Selected Economic Indicators in Uganda 1980-2017

Indicator                                                                    1980-89      1990-99     2000-09    2010-17

Budget balance, % of GDP-0.9  -2.6            -1.12        -3.8

Current Account, % of GDP -0.97            -4.3-3.1-8.3

External Debt, % of gross national income  40.2            74.4             43.7         29.6      

Lending Interest Rate                                                     25.8            17.7            20.3          22.6

Exchange Rate, Local currency per unit of U.S $        37.2           1026.4     1805.8      2832.9

Real GDP growth rate                                                   2.7             6.3                7.5            4.8

Inflation rate % change                                                  122.26       17.86.4          6.9

Source: Author’s computations using data from World bank (2017)

The budget and current account deficits are seen to be increasing in the period of 2000-2017, the real GDP declined from 6.3% in 1990-1999 to 4.8% in 2000-2017 while the inflation rate was also declining in the same period. These deficits hinder the nation’s development process and economic activities which affect real GDP growth (see Table 1) and future generations due to debt trap. Hence,thisneeds policy maker’s urgent intervention. There is a slight increase in lending interest rates (see Table 1) while exchange Rate of Local currency per unit of U.S dollars is increasing throughout the period. Uganda debt accumulated in the 1990’s as the government borrowed funds from financial institutions to boast current expenditure like finance public projects and strengthen the economy in order to fulfill IMF objectives adopted in 1980s. The external debt burden gradually reduced in the period of 2000-2017 (see Table 1) as Uganda is was one of the sub-Sahara countries  that benefited from the Heavily Indebted Poor Countries (HIPC) Enterprise enhanced in 1999 that emphasized debt relief in poor nations and this was later followed by Multilateral Debt Relief Initiative in 2005.

Research objectives.

The study aims to empirically analyze the validity of twin deficits in Uganda both in short and long run in the period of 1980-2017 and examine direction of causality between the trade deficit and budget deficit for policy implications.

1. Theoretical Framework and Empirical Literature.

Theoretical framework.

The short run link between fiscal and external deficit is analyzed in an open economy using the national income identity;

Y = C + I + G + (E – M)                                                                                       (1) 

where: Y is National income (GDP), C is Private consumption expenditure, I is investment expenditure, G is government expenditure, E is the export of goods and services, M is the import of goods and services hence (E-M) is as Current Account Balance (CA).

In an open economy savings(S) equals investment (I), the amount available for investment can go beyond domestic savings and if a country is incurring a negative current account that is importing more than exporting, equation (1) can be rearranged as;  

CA=Y-(C+I+G)(2)

The above equation can be rewritten as;

CA= S - I                                                                                                                       (3)

Where I= Y-C-G                                                                                                        (4)

Obtaining Total savings of a nation is by adding up the total of Private savings (Sp) and Government savings (Sg). Hence, Total Saving (S) is given by;

S= Sp+Sg (5)

Inserting private savings (Sp) and government savings in equation (3) gives;

 CA = Sg + Sp – I      (6)

Private savings is part of income after tax and can be expressed as;

Sp= (Y-T)-C                                                                                                                 (7)

Where, T=Taxes, Also, Government Saving is given by;

Sg= T-G (8)

Where G= Government spending

Substituting equation 7 and 8 in equation 3 gives; 

CA= (Sp-I) - (G-T)                                                                                                            (9)

The above formula indicates that current account balance (CA) depends on the savings deficit (Sp-I) while (G-T) is the fiscal deficit. The budget deficit is caused by the changes in private savings and investment. A rise in budget deficit lowers national savings hence negative current account balance. This in turn brings in the twin deficit hypothesis (Barro, 1974, 1989).The Keynesian analysis assumes that fiscal expansions reduce national savings and investment since consumers have a high marginal propensity to consume their incomes and this cause current account deficit. This deficit is further attributed to increase in government spending beyond the revenue, it upsurges domestic interest rates, and more capital inflows make the exchange rate to appreciate which in turn leads to a negative current account.  This appreciated exchange rate increase the consumption of imported goods (Mundell, 1963; Branson, 1976; Marston, 1985).

  The Ricardian equivalence hypothesis, REH (Barro; 1974, 1989)  disapproves the relationship between the external trade deficit and fiscal deficit, it assert that real interest rates is not affected by changes in taxes nor budget deficits and neither do increase in government expenditure affect current consumption and investment. He argues that individuals are rational to fiscal expansions. The country’s national saving remains unchanged as increase in private savings covers the low public savings and also used to pay future tax increases.

Related Empirical literature.

Like it has been studied by several economists, the increase in both fiscal and trade account balance are Marco-economic problems that many countries seek to solve and economists haveused various econometric methods to find out their cause and if there is a bi, unit-directional, reversecausal relationship between the two deficits. Some researchers have supported the Keynesian hypothesis, Twin divergence while others the Ricardian Equivalence Hypothesis (REH), below is some of the researchers analysis in different countries.

(Senadza1 &Aloryito, 2015) examined the twin deficit hypothesis (TDH) for Ghana in 1980-2014 and observed an insignificant short and long run co-integration among the deficits. Granger causality reveals a reverse causality that current account deficits cause budget deficits hence not supporting the TDH and this is common in commodity-based exporting countries (Alkswani, 2000).The worsening trade deficits negatively affect the country’s GDP which increase budget deficits.(Kahssay, 2018) investigated the twin deficit in Ethiopia in the period 1976-2015 using the VEC model and granger causality tests.Results showed a negative and statistically insignificant link between the trade deficit and budget deficit while granger causality test revealed the occurrence of a bi-directional causality between the deficits at 5% significant level. This causality result was similar to (Chaoneka, 2013) study in Botswana while Mozambique supported the Ricardian Equivalence Hypothesis (REH). Countries like Angola, South Africa and Seychelles had results that showed direct causal link from fiscal deficit to external deficit.

(Epaphra, M. 2017) examines the association between the two deficits in Tanzania using annual time series data for 1966-2015 and VEC Model. Empirical results support the Keynesian hypothesis and increase in fiscal deficits puts pressure on the current account deficitshence confirming the twin deficit hypothesis which is similar to the result of(Kim & Kim, 2006) for Korea, (Kearney &Monadjemi, 1990) for OECD countries and (Chaoneka, 2013) in Malawi and Zambia.Hence, there is need to adopt effect policies that reduce onthe budget deficits in each specific country. (Ekpenyong, 2014) used Granger Causality test and found out the internal balance in sub-Sahara Africa is as a result of improvements in the external sector. Hence the continent needs to balance trade by diversifying exports which will improve the fiscal balance.

(Njoroge et al, 2014) supported the mundell-flemingtheory; the study used the VAR techniques toexamine the TDHvalidity on Kenya using quarterly time series data beginning 1972 to 2012. The results indicate that there exists a long run relationship between the two deficits with variables of interest rates and exchange rates.

(Milne, 1977),  (Grier and Ye, 2009) supported the existence of long run linkbetween fiscal and current account deficits using OLS regressions on cross country data however  study did not account for  structural breaks in the variables in the time series data yet. (Mugume&Obwona, 1998) considered structural breaks while studying the deficits of African countries and reviewed that fiscal deficits have no effect on current account deficits in Uganda while (Lwanga& Mawejje, 2014) supported the Keynesian theory in Uganda in the long run using the VAR-VECM approach and concluded that current account deficits are as a result of increase in budget deficits and increasing interest rate.This result is similar to (Osama 2014) analysis in Egypt, (Ratha, 2011) in India and (Ganchev, 2010) in Bulgaria in the short-run while in the long-run period there exists no link between the two deficits in all countries.

(Wambui, 2016) used the VAR-GARCH to test the link between budget deficits and current account deficits in East African Countries Kenya, Uganda and Tanzania on time series data from 1980-2016. Bai and Perron Global Optimization approach was used to determine structural breaks and conditional heteroskedasticity in all countries. Results showed the existence of structural breaks in the countries with more in Uganda. Budget and current account balances had a positive and significant link in Kenya and Tanzania while the relationship was insignificant for Uganda.

(Ngakosso, A, 2016) does not support the Keynesian theory for the case study of the Republic of Congo, the study usedARDL approach to co-integration and observedthatcertainty current account balance is onlybetter when the budget deficit is combined in the analysis of economic policy to be implemented.

(Egwaikhide, 1997) analyzed the impact of fiscal deficit on trade deficit in the period of 1973-1997 in Nigeria since during that time the country experienced rapid trade deficits and other economic variables in the economy responded rapidly. Results reviled that fiscal deficits from increased expenditures were responsible for trade deficits and hence improvement in fiscal deficits was paramount for external balance in Nigeria. (Lloyd &Opeyemi, 2015) re-examined the long run co-integration between trade and budget deficits in Nigeria using a multivariate Granger causality within the VECM. Results revived that, its deteriorating current account that causes budget deficits and this calls for improvements in current account to solve the twin deficit problem. (Osama El-Baz, 2014) supported twin divergence hypothesis in Egypt, (Sakyi&Opoku, 2016) in Ghana and (Nguyen Van Bon, 2014) in 10 Asian countries.

Furthermore, many researchers have observed that current account deficits lower the growth of the economy causing fiscal deficits and this is termed as the unidirectional causality running from trade deficits to fiscal deficits. This is commonly found in small non-closed economies that depend on foreign direct investments (Baharumshah et al., 2006). (Mahuni, 2017) confirmed a long- run unidirectional reverse causality in Zambia using the VECM model. (Onafoora&Owoye2006) obtained a unidirectional causality and valid twin deficit in Nigeria and(Vyshnyak, O, 2000) obtained the same result in Ukraine.

In Pakistan, (Ateeq&Sumaira, 2017) used time series of 1972-2015 and the empirical study from co-integration techniques confirmed a one way causality running from external deficits to budget deficits in the long run while in the same country, (Tufail, M et al, 2014) used time series of 1972-2011 and observed a fiscal deficit has a positive effect on current account deficits in the long run and the two have bi-directional causality.

Most of the empirical literature has shown the link between the budget and the external deficits for countries in Africa with quite a few specific studies in Uganda. (Wambui, 2016) who analyzed the link between the two deficits in East African countries including Uganda did not consider the effect of other Marco-economic variables and direction of causality but confirmed a insignificant positive relationship between the two deficits in the long run.(Ahmad &Aworinde, 2015)confirmed a negative relationship between the deficits in Uganda while (Lwanga& Mawejje, 2014) confirmed a positive and significant uni-directional long run relationship running from budget to trade deficits using time series of 1999-2011 although, their study analyzed a shorter period according to the time series. And from 2010 to date the country has had widening fiscal deficits due to massive infrastructure developments which has affected other Marco-economic variables and economic growth in the country. Therefore, this study seeks to use a wider time series of 1980-2017 involving other Marco economic variables to examine the validity of twin deficits in the Uganda. TheVECM and pairwise Granger causality test is used in this case.

1. Model Specification, Data and Descriptive Statistics

The study uses annual time series data for the period 1980-2017 since during the mid-1980’s this is when the country had liberalization of trade of especially agricultural products and after mid 2000’s, the country had massive investments in infrastructure and other structural transformations that might be of great use in analyzing the twin deficit hypothesis. The data was collected from the World Bank data bases.The data of fiscal and current account balances was already provided as a percentage of GDP. The real effective exchange rate is used as one of the Marco economic control variable as it has an impact on trade balance and its measured as the value of a Uganda’s currency in relation to the average of othercurrencies divided by the index of costs. In case it appreciates, local products are less competitive in the outside market which affects different sectors in the economy. Real GDP further shows the economic performance of the Uganda while Lending interest rate has an effect on consumer spending and all these variables have a substantial effect on current account and international transfer of capital. The VEC Model estimation technique was used and data analyzed using E-VIEWS 8.

Therefore, this study estimated a model given as 

CA=f(FD,RGDP, r, REERI)which shows the functional form of the variables and this econometric model can also be written as;

CA(t)= α₀ + α₁ FD(t) + α₂ RGDP(t) + α₃ r(t) + α₄ REERI (t) + μ(t)                 (10)

Where CA: Current account; FD: Fiscal Deficit; RGDP: Real Gross Domestic Product; r: Lending interest rate; REERI: Real Effective Exchange rate Index, Ut: Stochastic error term; α₀,..α₄ : Regression coefficients and (t) is the time period .

Technique of Estimation

The model is estimated using Ordinary least squares regression technique to determine the statistical significance of the relationship between the variables. The econometric analysis of establishing the link between the fiscal/budget and current account deficits is done using the Granger causality (Amaghionyeodiwe et al 2015, Hiemstra& Jones 1994) and Vector Error Correlation model. This study tests for the long run and short run link and direction of causality between fiscal and current account deficit. The Unit root test is employed to understand the stationary variables and co-integration is used to know the long run link among variables in the model.

Description of Variables

Current Account (CA): This is the total sum of imports minus exports of goods and services adding net income from abroad. The current account deficit occurs when a country has insufficient funds to pay for foreign goods hence it will resort to borrowing. It is worse when current account deficit go beyond 3-4% of GDP. Current Account deficit is the dependent variable in the model of the twin deficit and expressed as a percentage of GDP.

Fiscal Deficit (FD): This is the excess of government spending over revenues earned.And in the model it’s expressed as a percentage of GDP.

Real Gross Domestic Product (RGDP):This is the measure of total worth of goods and services a country produces in a given period of time usually one year and adjusted to changes in prices. In the model it’s expressed in percentage form.

Interest Rate (r): This is the lending rate of the bank charged to the private sector as a result of short and medium loans obtained from the bank. Its normally given out for investment according to creditworthiness of borrowers. In the model is expressed in percentage form.

Real Effective Exchange Rate Index (REERI):This is the measure as the value of Uganda’s currency in relation to the average of other currencies divided by the index of costs.

 

Descriptive Statistics.

The descriptive statistics is carried out using E-Views 8 software to analyze the initial characteristics of the data sets. Table 2 below presents descriptive summary of the data.

The mean of current account (CA) was -4.003 with a minimum of -10.34 and maximum 2.86 while the mean of fiscal deficit (FD) was -2.02 with minimum -5.67 and maximum 0.38, both the mean of CA and FD are negatives which implies persistent deficits with in the period of 38 years.The Kurtosis value of RGDP, In REERI, In r, are greater than 3and Jarque-Bera test of RGDP, In REERI has probability value less than 5% which shows that they are not normality distributed in their frequency distribution of initial data, this is because there skewness values is different from the normal distribution value of zero.

 

 

 

 

 

 

Table 2; Descriptive summary of data

Variable

Obser.

Mean

Standard Deviation

Min

Max

Kurtosis

Jarque- Bera

Prob

CA

38

 

-4.0031

 

 

 3.0966

 

 

-10.340

 

 

2.8600

 

2.7996

1.3809

0.5013

FD

38

 

-2.0172

 

 

 1.5661

 

 

-5.6720

 

 

0.3880

 

2.3071

2.7113

0.2577

RGDP

38

 

 5.3868

 

 

 3.3489

 

 

-3.4000

 

 

10.400

 

4.0698

10.4202

0.0054

InREERI

38

 

 2.2462

 

 

0.3444

 

 

1.9740

 

 

3.3390

 

4.8821

21.5138

0.0000

In r

38

 

1.3531

 

 

0.1229

 

 

 1.0791

 

1.6020

3.2270

0.4885

0.7832

Source: Author’s estimations in E-Views 8.

Furthermore, Table 3 shows the correlation matrix which explains the magnitude of thelink between the variables. The results show that fiscal deficit (FD)and Real effective exchange rate index (In REERI) have apositive but weak correlation with the current account deficit of approximately 40 and 30percent respectively while other variables interest rate(r) and RGDP have a negative correlation with Current account.

 

Table3 : Correlation matrix

 

CA

FD

In r

In REERI

RGDP

CA

1.000

 

 

 

 

FD

0.399

1.000

 

 

 

In r

-0.108

-0.050

1.000

 

 

In REERI

0.291

0.416

-0.167

1.000

 

RGDP

-0.247

-0.057

-0.005

-0.474

1.000

Source: Author’s estimations in E-Views 8.

 

1. Analysis of Results

Unit Root Test

In order to test for stationary of time series, a unit root test is done using the Augmented Dickey-Fuller ADF (1979) Test to avoidthe generation of unauthentic results and ensuring proper estimation methods (see also Epaphra, M. 2017, Sakyi&Opoku, 2016).

The variables have a unit root is the null hypothesis andthis is not accepted if the unconditional value of the test statistic is greater than any critical values which implies stationary of variables.Also, there is non-stationary of variables if the null hypothesis is accepted.

According to table 4,results show that all variables are non-stationary at I(0) and after the transformation to first difference I(I) they become stationary.

 

Table 4; ADF unit root tests at Intercept & Trend, Lag length selection is based on Schwarz information criterion (SIC).

Variable

At Lag 0

Level of Significance

At Level(0)

At First Difference

CA

 

 

 

critical value

1%

-4.226

-4.234

 

5%

-3.536

-3.540

 

10%

-3.200

-3.202

 

Test statistics

-3.909

-9.008

 

 

FD

Critical Value

1%

-4.226

-4.234

 

5%

-3.536

-3.540

 

10%

-3.200

-3.254

 

          Test statistics

-2.636

-6.739

 

 

In R

Critical Value

1%

-4.226

-4.234

 

5%

-3.536

-3.540

 

10%

-3.200

-3.202

 

Test statistics

-2.537

-4.682

 

RGDP

 

Critical value

1%

-4.226

-4.234

 

5%

-3.536

-3.540

 

10%

-3.200

-3.202

 

Test statistics

-4.083

-6.538

 

 

In(REERI)

Critical value

1%

-4.226

-4.234

 

5%

-3.536

-3.540

 

10%

-3.200

-3.202

 

Test statistics

-3.664

-4.273

 

 

*MacKinnon (1996) one-sided p-values.

Null Hypothesis: there is a unit root

Source: Author’s estimations using E-Views 8.

 

 

 

 

Table 5: Lag Selection.

Lag

LogL

LR

FRE

AIC

SC

HQ

0

-178.2503

NA 

0.033045

10.77943

11.00389

10.85598

1

-79.27126

163.0242

0.000434

 6.427721

7.774510*

6.887015

2

-44.62374

46.87605*

0.000272

 5.860220

8.329333

6.702258

3

-18.07462

28.11083

0.000330

5.769095

9.360532

 6.993878

4

22.80512

31.26098

0.000244*

4.834993*

9.548753

6.442520*

*indicates lag order selected by the criterion.

Source: Author’s Computation, 2019.

 

The research used lag 4 for the model since FRE, AIC, HQ propose lag 4 at a 5% significance level as shown in Table 5 and this lag is used in running the johansen Co-integration results in table 6.

 

Johansen Co-integration Tests

In order to determine the long run link between fiscal and current account deficits plus other variables, the co-integration test is carried out using the trace and maximum Eigen value as suggested by (Johansen &Juselius, 1990). The advantage of Johansen technique isthat it strengthens the ADF stationary results (Teamrat, 2018) and estimates the long run equation giving the number of those that are co-integrating. The null hypothesis is no co-integration (r = 0) of models and this is rejected if both trace (λiz) and maximum statistics (λmx) are greater than the 5% critical value.

According to the output in Table 6, the alternative hypothesis is strongly accepted that there is co-integration of 4 models. The results suggest a stable long run relationship between theseries which implies rejection of the Ricardian equivalence hypothesis (REH) for Uganda in the long run however the validation of twin deficits in the country will depend on the direction of causality shown by the granger test and the corresponding statistical significance of the link. The maximum Eigen values indicate that some variables don’t deviate away from each other in the long run equilibrium path.

 

Table 6: Johansen Co integration test.

Sample (adjusted): 1985-2017

Included observations: 33 after adjustments

Series: CA, FD, In r, In REERI, RGDP

 

Number of                  Maximum             5%

Cointegrating            Statistic(λmx)          Critical               Trace                   5%

Ranks                                                  value             Statistic(λiz)           Critical value.

 

0                                       86.973              33.876              232.946            69.818

1                                       65.352              27.584              145.972             47.856

2                                       51.061              21.131               80.619              29.797

3                                       28.010              14.264               29.557              15.494

4                                       1.547                 3.841                 1.547                3.841

Source: Authors estimations in E-views.

 

Additionally, table 7 belowshows normalized co-integrating coefficients and the related standard

errorbelow each coefficient. The result indicate that in the long run, 1 percentage increase in fiscal deficit and Real effective exchange rate index will roughly lead to 2.046%  and 4.962% improvement  in current account deficit respectively ceteris peribus. Furthermore, 1 percent increase in lending interest rate and Real GDP worsenscurrent account deficit by 26.503% and 1.117% respectively and this is all significant at 5% level.

Table 7: Normalized co-integrating coefficients (standard error in parentheses)

     CA                   FD             In r              In REERI               RGDP               C

1.000          -2.046         26.503         -4.962                 1.117                 -32.240

1.  

Source: Authors computations from E-views 8.

 

VectorError Correction Model (VECM)

The error correction model is used to find out the long run stability of the current and budget deficitat first difference.  This model shows both the long run and short run disequilibrium or equilibrium in the variables (Senadza et al., 2016& Osama El-Baz, 2014).The variables are integrated at the same order, regression VECM results are summarized in Tables 8a) & 8b)of the main model when current account is the dependent variable and diagnostic tests are performed.  The value of R² is more than 60% which implies that the model is desirableandFigure 2 in appendix shows that the model satisfies the stability condition.  The (Breusch& Pagan, 1979) analysis of testing heteroscedasticity is done in table 12 in appendix section and the null hypothesis of no heteroscedasticity is not rejected since the probability of the chi-square is more than 0.05. Table 13 shows the Autocorrelation test for VECM model which indicates no evidence of serial correlation in all the 4 lags as the p-value is more than 5% significant level.

Table 8a) : Estimates of long-run Coefficients when CA is the dependent variable.

Variable

Long-run coefficients

T-statistic

FD

-2.046

-12.867

In r

26.503  

8.771

In REERI

-4.962

-3.111

RGDP

1.117

6.888

Constant

-32.240

-

ECT

-0.913

-2.461

Durbin-Watson stat        

2.343

R-Squared    

 

 

0.658

 

Source: authors estimates from E-views.

 

Table 8b): Estimates of Short run Coefficients when CA is the dependent variable

Variable

Short run coefficients

T-statistic

FD(-3)

-1.5157

-2.6338

RGDP(-1)

1.1143

2.7625

C

-0.355

-0.6549

Source: Authors estimates from E-views and only significant ones are selected.

Long run and short run estimates in table 8a) and 8b) above show that FD has a negative effect on CA and this is significant at 5% significant level.

Table 9: Wald test

Null Hypothesis: C(6)=C(7)=C(8)=C(9)=0from Eq(11)

(FD→ CA)

Test Statistic

Value

Df

Probability

 

F-statistic

 1.886072

(4, 11)

 0.1831

Chi-square

 7.544289

 4

 0.1098

 

Null Hypothesis: C(6)=C(7)=C(8)=C(9)=0from Eq(12)

(CA→ FD)

Test Statistic

Value

Df

Probability

 

F-statistic

 2.842361

(4, 11)

 0.0765

Chi-square

 11.36944

 4

 0.0227

 

Source: Authors estimates from E-views

According to the Wald test result in Table 9 above, there is no short run causality running from Fiscal deficit to current account deficit since the lagged values of FD are jointly statistically insignificant and therefore is no evidence of twin deficits hypothesis in the short run. On the other hand Wald results shows that there is short run causality running from CA to FD hence CA has fundamental effect on FD in Uganda since the lagged values of CA are jointly statistically significant at 5% level. Hence, there exists a uni-directional causality running from Current account deficits to Fiscal deficits in the short run.  As portrayed in Tables 8 a) and 8 b) above,  the coefficient of Fiscal deficit is negative both in short run and long run and hence fiscal deficit is found to have a negative relationship with current account and statistically significant at 5% significant level.

These results support a Twin Divergence Hypothesis(TDivH)in Uganda and opposing firstly, the Ricardian Equivalence proposition that postulates that Fiscal deficit has no effect current account deficitand secondly the Keynesian Hypothesis that claims a positive link between CA and FD. The (TDivH)implies that budget deficits can either worsen or improve trade deficits or trade deficits can either worsen or improve budget deficits. In economic recessions associated with reductions aggregate demand and investment, fiscal deficits increase and current account is improved while in economic booms,aggregate demand increases the fiscal deficit and with improved investment which worsens trade deficits (sees also Baxter, 1995,Kim &Roubini, 2008, Van Bon, N. 2014).

Finding show that increase in budget deficits improve trade deficits however this does not signify that the government should increase its expenses to improve trade deficits. In factin this scenario, it needs to take caution about any policy that might increase fiscal deficits especially during unstable economic periods. Therefore public expenditure should be in development projects that increase employment. According to (Sakyi&Opoku, 2016) the government should increase expenditures only in short run periods in productive sectors and lowering taxes for private export oriented firms. However much this can make it run a deficit but at the same time it will boast exports and encourage more private sector investment plus improved people’s incomes.

The estimates of the VECM model in Table 11 in the appendix when CA is the dependent variable show that the probability of the target variable C(1) which is the speed of adjustment or error correlation co-efficient is negative and significant at 5% level which implies long run convergence of about 91%disequilibrium in CA obtained each period. There is a stable long run influence from the Independent variables to the dependent variable which is the current account.

Granger Causality Test.

 

Table 10: Pairwise Granger Causality test.

 

 

Null Hypothesis (H0)

 

Observations

 

F-statistic

 

 

Probability

 

Decision

 

 

 FD does not Granger Cause CA

 

34

1.16877

0.3483

Accept Ho

CA does not Granger Cause FD

34

5.39374

0.0028

Reject Ho

Inr does not Granger Cause CA

34

0.2009

0.9355

Accept Ho

CA does not Granger Cause Inr

34

 

 0.76579

 

 

0.5504

 

Accept Ho

In REERI does not Granger Cause CA

34

 

 

0.96593

 

 

 

0.4435

 

Accept Ho

CA does not Granger Cause In REERI

34

 

 0.77728

 

 

0.5504

 

Accept Ho

RGDP does not Granger Cause CA

34

 

 0.37311

 

 

0.8255

 

Accept Ho

CA does not Granger Cause RGDP

34

 

 0.35424

 

 

0.8386

 

Accept Ho

Source: Authors estimates in E-Views.

 

Different literature has discussed the link and dynamics between the current and fiscal deficits and many have come up with contradicting results due to the different statistical methods and time periods taken. For effective policy purposes the direction of causality is determined in line with previous studies and the Granger-Causality result in table 10 still indicates auni-directional causality. CA is granger causing FD since the p-value is less than 5% and the null hypothesis of no causality is rejected, supporting the reverse causality proposition. Hence for the Ugandan economy, there is a revered causality and this implies that fiscal expenditure responds strongly to current account changes and if current account deficit is improved, fiscal expenditure reduces and this leads to improvement in fiscal deficits. It is also observed that none of the individual variables granger causes current account.

Theexistence of twin divergence in Uganda is contrary to majority of the studies done in Uganda and other countries, for example (Mugume&Obwona, 1998) who disqualified the link between the two deficits in Uganda,(Lwanga& Mawejje, 2014) confirmed the Keynesian hypothesis in Uganda with budget deficits causing trade deficits while (Wambui, 2016) confirmed a positive insignificant link between the deficits. The results are similar to (Ahmad &Aworinde, 2015) who confirmed a negative link between the deficits in Uganda, (Osama El-Baz, 2014) who found reverse causality and twin divergence in Egypt, (Sakyi&Opoku, 2016) in Ghana, Javid et al. 2010 for Pakistan, Nguyen Van Bon (2014) in 10 Asian countries. The causalityresults collaborate with the analysis of (Bernardin et al., 2016) in Ghana, (Sobrino, 2013) in Peru, (Egwaikhide et al., 2002), (Lloyd, 2015) for Nigeria, (Suchismita&Sudiptal, 2011) for India among otherswho confirm a uni-directional causality running from external deficits to budget deficits.

However much the granger causalitytest revealsthat Real GDP growth and current account deficits are not related in Uganda, the coefficient of RGDP is positive and hence increase in RGDP worsens Current account and this is significant at 5% level. This might be caused by increased economic activities due to Uganda’s growing private sector that employ different peopleand improve their realincomes which boast demand. Theincrease in demand formore industrialized capital and consumer goods which are imported like motor vehicles, construction materials, alcoholic products among othersincrease import invoices and hence atrade deficit (see also Epaphra, M. 2017, Bernardin et al 2016). It is also most expected that increase in RGDP is associated with expansion in production levels which improves the export volume hence improving current account deficits. 

Findings also show that the lagged interest rate coefficients are negative and insignificant. This implies that increase in this interest rate does not necessarily improve CA deficits. In fact in the long run, increase in interest rate worsens current account deficits. Uganda’s lending interest rates have been high as the Bank of Uganda is trying to achieve its policy priority of curbing inflation, strengthen exchange rates and allowing banks to extend credit to more risky projects.However interest rate impact on trade deficits is most times un predictable. Trade deficits worsen with increase in interest rate if the exchange rate has appreciated making imports cheaper leading to high import invoices and low export receipts. However this is contrary to Uganda’s case because the Ugandan Shillings has often depreciated as imports are increasing and the lagged exchange rate is insignificant. Infact (Bawumia, 2014) and (Sakyi&Opoku, 2016) observed the same scenario in Ghana. Despite the country’s urge to increase interest rate and solve some economic problems, in the process people are discouraged from borrowing to investsince it becomes expensive and there is also increase in price of goods and services (see Kwakye, 2010) causing low consumption levels.Since most consumed goods in Uganda are imported, low consumption levels will result in improved trade deficits hence a contrary link between trade deficit and interest rates.

1. Summary of Findings, Policy Suggestions, Conclusion.

The study sought to empirically analyze the validity of twin deficits in Uganda both in short and long run in the period of 1980-2017 and examine direction of causality between the trade deficit and budget deficit since there was a persistent co-movement of both deficits over the past three decades. Augmented Dickey-Fuller ADF (1979) test is performed on time series and all variables are stationary at first difference. Empirical VECM results have shown that considered Marco-economic variables current account deficits, fiscal deficits, real GDP,Real effective exchange rate index and lending interest rate are co-integrated. The speed of adjustment is negative and significant which indicates a steady long run link among variables. The VECM confirms a statistically significant negative link between budget deficits and current account deficits both in short run and long run which indicates evidence of Twin divergence hypothesisfor Uganda. This implies that budget deficits can either improve or worsen trade deficits or trade deficits can either worsen or improve budget deficits. The findings show that increase in Fiscal deficits improves current account however not necessarily meaning that government should increase its expenses to improve trade deficits as this might negatively affect the economy especially in unstable economic periods. That is why granger causality test helps in determining the direction of causality for better policy implication. The results show a unidirectional causal link which is reversed running from current account deficits to fiscal deficits.

This is due to trade openness of Ugandan economy,increasing the country’s import invoices much higher than the export receipts. It is exposed to external price shocks and the exportation of a limited variety plus low valued agricultural products that are out competed or attract less prices on external market. Policy initiatives should be directed to supervising current account deficit like increasing production and value addition to Agriculture products exported in order to increase foreign exchange earnings. Maintaining a conducive Marco-economic environment with reduced inflation rates, improved real exchange rate,improving domestic infrastructure inform of roads, electricity supply, proper storage facilities for agriculture products will boast domestic production, expand employmentand minimize on massive importation of consumer goods. In all these, the governmentdoes not only collect more revenues to cover the budget deficitbut alsotrade deficits are lowered.

The Ugandan government needs to adopt to import substitution and export promotion industrial strategies that will further boast domestic firm production, this also saves that country from imported inflation, and increase earnings of people. The findings suggest that government should strengthen external policiesand if well implemented will improve foreign earnings, national savings hence improved current account balance which reduces fiscal deficits.Lastly the government of Uganda needs to minimize on excessive non-development expenditures especially in  political administrations that in most cases over consume tax payers money with no or less profit.

 

[1]https://www.theglobaleconomy.com/Uganda/trade_openness.

 

[2]https://www.theeastafrican.co.ke/business/uganda-budget-focus-on-infrastructure-does-not-bear-fruit-/2560-3875812-29eigoz/index.html

 

[3]https://ugandabankers.org/here-is-why-the-uganda-shilling-is-weakening/

 

[4] These debt relief programs were supported by the IMF under the program of Heavy Indebted Poor Countries(HIPC) which totaled up to US$ 2billion and other poverty reduction programs.