Professional Documents
Culture Documents
Examining Twin Deficit Hypothesis Using Simultaneous Model in Pakistan
Examining Twin Deficit Hypothesis Using Simultaneous Model in Pakistan
Examining Twin Deficit Hypothesis Using Simultaneous Model in Pakistan
Ismat Nasim
Lecturer, Department of Economics
Government Sadiq College Women University, Bahawalpur
Email: ismat.nasim@gscwu.edu.pk
Kashif Raza
Lecturer, Department of Economics
The Islamia University of Bahawalpur, Bahawalnagar Campus
Email: kashif.raza@iub.edu.pk
Abstract:
This research is an attempt to ferret out dynamics of twin deficit
hypothesis in Pakistan using annual data from 1974 to 2018, covering
45 years. To address empirical association between fiscal deficit and
current account deficit, simultaneous model consisting of two
stochastic equations is utilized. Hausman’s simultaneity test is utilized
to check endogeneity of the variables and stationarity of the variables
is confirmed through application of Ng-Perron unit root test. 3SLS
technique is used to empirically estimate the model. Our empirical
results strongly support significant positive two way association
between fiscal deficit and current account deficit. The key policy
prescription is that there is dire urgency of coordinated and prudent
monetary and fiscal policy to overcome pitfalls of rising twin deficits in
Pakistan.
I. Introduction
The purported relation of twin deficit hypothesis energized extensive empirical
and academic debate especially during 1980’s when most of the countries around the
globe faced severe crises of escalated fiscal deficit and deteriorated current account
balance simultaneously due to International oil price hikes of 1970’s beside other
potential determinants. However, twin deficit literature traces back to pioneered work of
Fleming (1967) that shows positive linkages between fiscal deficit and current account
deficit, for details see, Fleming (1968) and Abell (1990).
Another significant rationale of twin deficit hypothesis can be explained with the help of
open economy model of Keynes. According to this, increase in budget deficit results in
rise of total demand which in turn causes rise in domestic interest rates. The increased
domestic interest rate attracts foreign investment which can substantially appreciate
domestic currency and makes exports expensive; imports cheap. Thus negatively affects
net exports and results in trade deficit or current account deficit. Proponents of Keynes
preposition includes Darrat (1988) and Bachman (1992) among others.
Recent advances in twin deficit hypothesis empirical literature came up with two
more conclusions on subject matter i.e. 1). There is reverse causality running from
current account deficit to fiscal deficit. 2). There is bi-directional relationship between
variables. The concept of reverse causality is mainly holds in case of small open
developing countries that excessively relied on external capital inflows to augment their
scarcity of finance and resources. The proponents of reverse causality includes Summer
(1988), Hatemi and Shukur, (2002) among others. The proponents of bi-directional
relationship levied severe criticisms on using of single equation framework to investigate
twin deficit hypothesis as results of single equation model suffers from simultaneity bias.
Table 1: Decade Wise Analysis of Fiscal Balance and Current Account Balance
(1970-2018)
Decade FB/GDP (%)1 CAB/GDP (%)2
1970’s -8.05 -4.6
1980’s -6.73 -3.13
1990’s -7.15 -3.16
2001-05 -3.56 1.61 (Surplus)
2006-10 -6.46 -4.78
2011-15 -6.86 -1.30
2016-18 -5.63 -4.62
Note: Author’s calculations
1
Fiscal Balance as percent of GDP.
2
Current Account Balance as percent of GDP.
1619
in table 1 strongly support that there is strong link between fiscal balance and current
account balance as periods of high fiscal gap are associated with periods of high current
account gap.
Therefore, rising twin deficit of Pakistan calls researchers to come forward and
to undertake in-depth analysis of subject matter. The primary objective of this paper is to
contribute in empirical literature through joint estimation of twin deficit dynamics in
Pakistan over longer time period. This research significantly differs from past empirical
research done on subject matter. Firstly, it addresses deficiency in econometric
methodology by adopting simultaneous structural model instead of relying on a single
stochastic equation model. Secondly, it also explores direct and indirect effects of all
exogenous variables. Thirdly, full information method i.e. 3SLS is utilized as an
estimation strategy after ensuring its pre-requisites.
With regard to reverse causation i.e. current account deficit causes budget
deficit, there exists number of empirical evidences. Few of them includes Khalid and
Teo, (1999), Alkswani (2000). The bi-directional relationship between variables is
reported in work of Normandin (1999) among others. A review of empirical evidences on
subject matter during 1990-2020 is reported in tabular form in order to have quick bird’s
eye view.
1620 Pakistan Journal of Social Sciences Vol. 40, No. 4
3
Current Account Deficit
4
Budget Deficit
5
Error Correction Model
6
Unrestricted Error Correction Model.
7
For Malaysia data ended at 1998: Q2 before pegging of its currency to US dollar.
8
Vector autoregressive
1621
causality
Ravinthirakumaran et al. 1980-2012 Co-integration, Pakistan & Sri Lanka:
(2016) (Annual data) ECM, Granger BD → CAD
SAARC Countries Causality Nepal, India &
(Pakistan, Bangladesh, Bangladesh: CAD →
India, Sri Lanka and BD
Nepal)
Bhat and Sharma, (2018) 1970 – 2016 Non-linear ARDL9 Confirms association
India (Annual data) between CAD and BD
but instead of linear
association, study
confirms asymmetric
relationship between
CAD and BD.
Manzoor et al. (2019) 1973-2017 VAR Trade deficit directly
Pakistan (Annual data) IRF10 causes budget deficit,
Granger Causality whereas, budget
deficit causes trade
deficit though different
channels.
Shruti (2019) 1985 – 2016 ARDL and Toda Pakistan & Sri Lanka
South Asian Countries (Annual data) Yamamoto (CAD → BD)
(Pakistan, India, Causality Nepal (BD → CAD)
Bangladesh, Nepal, Sri (for time series India & Bangladesh
Lanka) analysis) (bi-directional
GM – FMOLS11 association between
GM – DOLS12 CAD and BD)
(for panel data
analysis)
Bucevska (2020) 2005:QI –2017:QIV VAR Long run relationship
Republic of North (Quarterly data) Granger Causality exists between CAD
Macedonia VECM and BD.
Note: → denotes one way causation
From reported work on subject matter we can safely conclude that literature
remains inconclusive. Thus it demands more research on subject matter. To best of our
knowledge there is no research that have utilized simultaneous model in context of
Pakistan. In order to bridge this gap in literature, this research is undertaken.
(a)
9
Auto –Regressive Distributed Lag Model
10
Impulse Response Function
11
Group Mean – Fully Modified Ordinary Least Square
12
Group Mean – Dynamic Ordinary Least square
13
Net exports
1622 Pakistan Journal of Social Sciences Vol. 40, No. 4
(b)
Since we know that national savings can be bifurcated into two distinct
categories i.e. private and public savings therefore algebraically we can write it as:
(e)
Where, P.Sav (private savings) = Disposable income (Yd)14 minus Consumption
(Cons), G.Sav (government savings) = Revenue (TX)15 – Expenditures (G+TR)16. The
equation of national savings can also be expressed algebraically as:
Keeping in view equation (d) and (e’), we can state eq (f) as follows:
(f)
14
Y-TX (income minus taxes)
15
Taxes (direct + indirect)
16
Expenditures consist of government expenditures and transfer payments (e.g. old age allowances,
unemployment allowances etc.)
1623
In order to show effect of fiscal deficit on current account balance, we can
rewrite above equation as:
( )
Or,
(h//)
The above final equation stipulates twin deficit conventional view that if (P.Sav
– Inv) is fixed or constant an increase in government expenditures (keeping tax
collections remain unchanged) positively affects current account deficit. This view is
advocated by Keynes and his proponents.
A. Model Specification
Following structural equations are formulated to investigate twin deficit
interactions in Pakistan.
17
We cannot take log of current account deficit even after taking as percent of GDP because there exist three
values whose log is non negative.
1624 Pakistan Journal of Social Sciences Vol. 40, No. 4
B. Research Methodology
Time series data often exhibit trending pattern therefore to avoid spurious
regression results we need to check stationarity and for this we have used Ng-Perron unit
root test (2001) which is the most powerful and recommended test among others because
it overcomes the issues related to poor size and power distortions of ADF test and PP test.
In second stage, we estimate all structural parameters with the aid of OLS
method. It is relevant to mention here that before estimation of structural
parameters, we have to replace values of actual regressands with respective
fitted values of regressands.
Over identification of the model is confirmed through order and rank condition
of identification. The correct specification of the model is checked by Ramsay RESET
test. Serial correlation of residual terms is checked by Breusch-Pagan-Godfrey Serial
Correlation test.
1625
V. Results and Discussions
A. Stationarity Results
After checking stationarity of the data, pre-conditions of the 3SLS method are
confirmed through appropriate statistical tests as mentioned in section IV.
Table 4: Ng-Perron Unit Root Test Results 1974-2018
MZα MZt MSB MPT
Deterministic Deterministic Deterministic Deterministic
Variables terms terms terms terms
c c,t c c,t c c,t c c,t
Ng-Perron in Levels
CAD/GDP -9.8** -10.4 -2.1** -2.2 0.22** 0.21 2.6** 8.7
ln (FD/GDP) -10.3** -11.7 -2.2** -2.3 0.21** 0.20 2.3** 7.9
ln EXR 0.03 -5.1 0.01 -1.5 0.54 0.30 21.4 17.5
ln TOT -5.1 -8.2 -1.2 -1.9 0.24^ 0.23 5.5 11.3
ln (ED/GDP) -5.7 -7.2 -1.6 -1.8 0.29 0.26 4.2^ 12.5
ln (M2/GDP) -5.1 -23.2** -1.3 -3.3** 0.26^ 0.14** 5.3 4.0*
Ng-Perron in First Differences
dCAD/GDP -17.4* -17.4** -2.9* -2.9^ 0.16* 0.16** 1.4* 5.2**
dln(FD/GDP) -15.4* -16.5^ -2.7* -2.8^ 0.18** 0.17^ 1.5* 5.7**
dlnEXR -17.2* -17.3^ -2.9* -2.9^ 0.17** 0.16** 1.4* 5.2^
dlnTOT -1.2 -15.0^ -0.7 -2.6 0.57 0.18 17.3 6.5^
dln(ED/GDP) -1.7 -16.8^ -0.9 -2.8^ 0.52 0.17^ 13.7 5.5^
dln(M2/GDP) -12.5** -20.8** -2.3** -3.2** 0.19** 0.15** 2.4** 4.4**
Critical Values
1% -13.8 -23.8 -2.58 -3.42 0.17 0.14 1.78 4.03
5% -8.10 -17.3 -1.98 -2.91 0.23 0.17 3.17 5.48
10% -5.70 -14.2 -1.62 -2.62 0.27 0.18 4.45 6.67
Note: * is used for significance at 1 %, ** at 5 % and ^ at 10 %, “c”→ constant, “c,t” → constant and trend,
MZα → Modified Philips-Peron test, MZt → Modified PP t-test = MZα * MSB, MSB → Modified Sargan-
Bhargava test, MPT → Modified Point Optimal test.
In current account deficit equation three variables out of four are statistically
significant i.e. fiscal deficit as percent of GDP, nominal exchange rate and per capita
income. Fiscal deficit has significant positive impact on current account deficit with a
reported coefficient 5.85. The estimated coefficient can be precisely interpreted as 1
percent increase in fiscal deficit, on average, directly endorse or increases current account
deficit by 5.85 units in the long run, given other factors constant. This empirical finding
supports conventional view of twin deficit hypothesis that fiscal deficit causes current
account deficit by escalating trade deficit and it is also consistent with current experience
of Pakistan’s economy and with findings of Piersanti (2000), Cavallo (2005) and Iram et
al. (2011) on subject matter. The rising trend of Pakistan’s fiscal deficit has caused
1626 Pakistan Journal of Social Sciences Vol. 40, No. 4
serious impediments for current account balance as increased fiscal deficit forced
government to borrow externally as well as internally which has adversely impacted on
current account balance.
With regard to current account deficit and exchange rate relationship we came
to know that nominal exchange rate has exerted positive impact on current account deficit
and it is significant in Pakistan. The estimated coefficients can be precisely interpreted as
1 percent increase in nominal exchange rate directly, on average, increases current
account deficit by 2.90 units in the long run. Since the establishment of Pakistan, it has
faced severe depreciation of exchange rate as 9.9 rupees per US $ in 1974 to 68.28 rupees
per US $ in 2008 to 84.56 rupees per US $ in 2010 and to 109 rupees per US $ in 2017.
Because of this massive depreciation Pakistan’s currency is becoming weaker and attracts
less foreign investment or foreign capital and causes current account deficit. With regard
to current account deficit and terms of trade nexus, we found negative insignificant
association between the variables. Due to insignificance of the variable no valid
conclusion can be drawn.
With regard to current account deficit and per capita income nexus, we found
significant and negative association between the variables. In this research we have used
per capita income as proxy for economic growth. Higher economic growth means lesser
current account deficit which is theoretically valid. The empirical estimate of the variable
(economic growth) indicates that 1 percent increase in economic growth, on average,
directly can lowers current account deficit by 6.86 units in the long run. This finding has
an important policy implication that Government should focus on channels of economic
1627
growth and put efforts to achieve past growth momentum of 1960’s in order to curb
current account deficit and to achieve financial competitiveness.
In fiscal deficit equation all four variables are statistically significant and
carrying expected signs. We also find significant two-way relationship between fiscal
deficit and current account deficit. The increase in fiscal deficit positively affects current
account deficit or vice versa. The estimated coefficient of current account deficit is
interpreted as 1 unit increase in current account deficit directly, on average, increase
fiscal deficit by 0.10 percent in the long run. The rationale behind this empirical finding
is that because mostly Pakistan’s current account deficit was financed by borrowings
(internal or external) which caused huge government debt. Due to accumulated debt,
interest payments have drastically increased over the years and Pakistan has trapped into
larger persistent fiscal deficits. Furthermore, composition and structure of foreign
assistance has changed from grants to hard term loans which are detrimental to economic
growth and current account balance.
With respect to current account deficit equation, the direct and total impact of
nominal exchange rate on current account deficit is showing same sign although size of
the respective effect is quite different. The positive sign implying that more nominal
1628 Pakistan Journal of Social Sciences Vol. 40, No. 4
exchange rate in turn causes more current account deficit or lower nominal exchange rate
in turn results in lower current account deficit. We can easily compute magnitude of
indirect effect as 6.35 (total effect) minus 2.90 (direct effect) is equal to 3.45 (indirect
effect) which reveals that indirect impacts of exchange rate on current account deficit
outstrip its direct effects on current account deficit. Similarly, direct and indirect effects
of per capita income on current account deficit are also negative indicating economic
growth dampens current account deficit through both direct and indirect channels.
With respect to fiscal deficit equation, the direct impact of external debt as
percent of GDP on fiscal deficit is less than total impact as direct effect accounts for 0.91
percent increase in fiscal deficit whereas total impact accounts for 2.08 percent increase
in fiscal deficit given that 1 percent increase in external debt to GDP ratio. Therefore we
conclude indirect impact of debt through debt servicing channel is more important than
actual debt stock as percent of GDP. The adverse effects of debt servicing can be
visualized as it absorbs country’s own resources or revenues. The indirect impact of
money supply on fiscal deficit is not statistically significant therefore valid conclusions
cannot be drawn with precision.
VI. Conclusion
This study empirically explores simultaneous relationship between current
account deficit and budget deficit in Pakistan using annual observations from 1974-2018.
This research is of great significance given rising trend of twin deficit in Pakistan. After
ensuring stationary properties of the data, 3SLS method is utilized to estimate the
structural model on subject matter. Total and direct effects are also reported. The 3SLS
empirical results confirmed statistical significance of most of the variables used in the
analysis of twin deficit hypothesis. Secondly, empirical results are aligned with
theoretical expectations and also consistent with existing situation of the economy. The
core empirical finding of the study is that it strongly supports two-way relationship
between budget deficit and current account deficit during the investigated period. To
overcome ill effects of rising twin deficit following suggestions are made in the light of
estimated model.
Hatemi, A. & Shukur, G. (2002). Multivariate Based Causality Test of Twin Deficits in
US. Journal of Applied Statistics, 29, 817-824.
Iram, M. P. S., Shadid, A., Mahpara, S., & Fazli, R. (2011). Old wine in new bottles:
testing the Keynesian preposition of twin deficit in case of
Pakistan. International Journal of Business and Social Science, 2(5), 209.
Kaufmann, S., Scharler J. & Winckler G. (2002). The Austrian Current Account Deficit:
Driven by Twin Deficits or by Inter temporal Expenditure Allocation. Empirical
Economics, 27, 529-42.
Khalid, A.M & W.G Teo (1999). Causality Test of Budget and Current Account Deficits:
Cross Country Comparisons. Empirical Economics, 24, 389-402.
Khan, A.M & T.W Guan (1999). Causality Tests of Budget and Current account Deficits:
Cross Country Comparisons. Empirical Economics, 24(3), 389-402.
Kim, K,H. (1995). On the Long Run Determinants of US Trade Balance: A Comment.
Journal of Post Keynesian Risk Sharing. Working Paper.
Laney, L. (1984). The Strong dollar, The Current Account and Federal Deficit: Cause and
Effects. Federal Reserve Bank of Dollars Economic Review, 1-14.
Manzoor, H., Younas, M. Z., Mahmood, R., and Rizwan, M. A. (2019). A Twin Deficit
Hypothesis: The Case Study of Pakistan. Bulletin of Business and Economics,
8(3), 117-131.
Miller, S. M & Russek, F.S. (1989). Are the Twin Deficits Really Related?
Contemporary Policy Issues, 7, 91-115.
Nishat, M. & Aqeel, A. (2000). The Twin Deficit Phenomenon: Evidence from Pakistan.
The PDR, 39(4), 535-550.
Piersanti, G. (2000). Current Account Dynamics and Expected Future Budget Deficits:
Some International Evidence. Journal of International Money and Finance, 19.
Ravinthirakumaran, N., Selvanathan, S. and Selvanathan, E.A. (2016). The twin deficits
hypothesis in the SAARC countries: an empirical investigation. Journal of the
Asia Pacific Economy, 21(1), 77-90.
Summers, L. (1988). Tax Policy and international Competitiveness. J.A Frankel (ed)
International Aspects of Fiscal Policies (NBER) Conference Report, Chicago
University Press.
Yasmin, F. (2015). Twin Deficit Hypothesis: A Case of Pakistan. Sukkur IBA Journal of
Management and Business, 2(1), 71-84.
Copyright of Pakistan Journal of Social Sciences (PJSS) is the property of Bahauddin
Zakariya University and its content may not be copied or emailed to multiple sites or posted
to a listserv without the copyright holder's express written permission. However, users may
print, download, or email articles for individual use.