EXCHANGE RATE VOLATILITY AND ITS EFFECTS ON MACROECONOMIC MANAGEMENT IN ZAMBIA Jonathan M. Chipili * Final Draft Report December 2014 Abstract The impact of exchange rate volatility on selected macroeonomic indicators in Zambia is analysed over the period 2007-2013. The extent of exchange rate volatility deemed disruptive on the economy is also determined. A significant role for exchange rate volatility in macroeconnomic management is established: volatility in the kwacha/US dollar exchange rate tends to depress trade, increase inflation, discourage capital flows and dampen output although its impact tends to be temporary; and exchange rate depreciation in excess of 1% tends to generate negative effects on the economy. The results underscore the significance of the exchange rate in trade and monetary policies and the transimission of exchange rate impulses to the rest of the economy. The central bank must therefore seek to stabilize the exchange rate; however, this requires an appropriate currency stabilization strategy that ensures currency flexibility and competitiveness. Interventions in foreign exchange markets should not be undertaken in isolation as this may significantly deteriorate the reserves position and weaken the country’s ability to respond to adverse external shocks. Thus, complementary policies are recommended to sustain a stable exchange rate. JEL classification: F31; C22; C32 Keywords: Exchange rate volatility, macroeconomic indicators, GARCH, SVAR * Bank of Zambia, P.O. Box 30080, Lusaka, 10101. Zambia. E-mail: [email protected]Tel: +260 211 228888-92. The views expressed in this paper do not in any way represent the official position of the Bank of Zambia. I remain responsible for all the errors and omissions. 1
Transcript
EXCHANGE RATE VOLATILITY AND ITS EFFECTS ON MACROECONOMIC MANAGEMENT IN ZAMBIA
Jonathan M. Chipili*
Final Draft ReportDecember 2014
Abstract
The impact of exchange rate volatility on selected macroeonomic indicators in Zambia is analysed over the period 2007-2013. The extent of exchange rate volatility deemed disruptive on the economy is also determined. A significant role for exchange rate volatility in macroeconnomic management is established: volatility in the kwacha/US dollar exchange rate tends to depress trade, increase inflation, discourage capital flows and dampen output although its impact tends to be temporary; and exchange rate depreciation in excess of 1% tends to generate negative effects on the economy. The results underscore the significance of the exchange rate in trade and monetary policies and the transimission of exchange rate impulses to the rest of the economy. The central bank must therefore seek to stabilize the exchange rate; however, this requires an appropriate currency stabilization strategy that ensures currency flexibility and competitiveness. Interventions in foreign exchange markets should not be undertaken in isolation as this may significantly deteriorate the reserves position and weaken the country’s ability to respond to adverse external shocks. Thus, complementary policies are recommended to sustain a stable exchange rate.
* Bank of Zambia, P.O. Box 30080, Lusaka, 10101. Zambia. E-mail: [email protected] Tel: +260 211 228888-92. The views expressed in this paper do not in any way represent the official position of the Bank of Zambia. I remain responsible for all the errors and omissions.
The study of persistent variability in the exchange rate post-1973 remains an active subject of empirical investigation due to the adverse effects a volatile exchange rate can potentially impose on the economy. Policymakers are interested in exchange rate volatility for formulating suitable macroeconomic policies while corporate entities are concerned about its implications on profits by designing appropriate risk management tools (Bollen, 2008; Bauwens and Sucarrat, 2006).
Aggregate and sectoral output, price level, international trade and foreign investment are typically affected by a volatile exchange rate (Mirdala, 2012; Arratibel, et al., 2011; Takagi and Shi, 2011; Chipili, 2010). Volatility in the exchange rate introduces uncertainty which in turn generates negative economic welfare effects (Bergin, 2004; Obstfeld and Rogoff, 1998). Further, fluctuations in the exchange rate affect consumer goods prices which in turn affect demand and consequently consumption. Monetary policy is also affected by currency fluctuations especially where domestic growth is underpinned by exports as authorities attempt to support the external sector through exchange rate stabilisation at the expense of inflation stabilisation (Crosby, 2000). Echange rate uncertainty can also create incentives for trade protectionist tendencies and sharp currency reversals which in turn impose further costs on the economy (Sengupta, 2002; Bayoumi and Eichengreen, 1998).
Volatility in exchange rates can also restrict the flow of international capital by reducing direct and portfolio investments. Speculative capital flows may also be induced by exchange rate volatility under the flexible regime that could in turn contribute to instability in economic conditions (Willett, 1982). Further, greater exchange rate volatility increases uncertainty over the return of a given investment. Potential investors are attracted to invest in a foreign location as long as the expected returns are high enough to compensate for the currency risk. In view of this, foreign direct investment tends to be lower under higher exchange rate volatility. Most developing economies are net debtors such that considerable fluctuations in exchange rates may affect the real cost of debt service. Similarly, banks, corporate entities and individuals exposed to foreign denominated debt are negatively affected by exchange rate volatility (Schnabl, 2009).
Central banks globally therefore endeavour to stabilize exchange rates in order to moderate the adjustment and uncertainty costs that a volatile exchange rate imposes on the economy.
While a vast amount of literature on the impact of a country’s exchange rate volatility on macroeconomic variables exists, limited empirical work has been undertaken on Zambia. Notably, Chipili (2013), Musonda (2008) and Tenreyro (2007) have examined the exchange rate volatility-trade link; Aghion et al. (2009) have analysed the impact of exchange rate volatility on productivity growth; and Chipili (2013) examined the relevance of exchange rate volatility in monetary policy setting as well as central bank foreign exchange intervention decision. In all these studies, the revelance of exchange rate volatility is noted with varying degree of impact on the sectors/variables studied. While a number of studies report negligible exchange rate volatility effects on real economic activity, Aghion et al. (2009) find exchange rate volatility to exert significant negative impact on productivity growth subject to a country’s level of financial development: the impact is severe in thin and less developed capital markets.
This study seeks to empirically examine the impact of the kwacha exchange rate volatility on selected macroeconomic variables in Zambia over the period 2007-2013 due to data limitations especially on capital flows at monthly interval. A brief perspective on the macroeconomic policy evolution from early 1990 when economic reforms commenced is provided. Specifically, the study analyses trends in the kwacha/US dollar exchange rate and
2
major macroeconomic variables. Possible causes of exchange rate volatility in the kwacha exchange rate are reviewed from past studies. A formal econometric framework for examining the impact of the kwacha/US dollar exchange rate volatility on the economy and the threshold for exchange rate volatility deemed disruptive to the economy are provided.
The results reveal that exchange rate volatility tends to depress trade, increase inflation, discourage short-term capital flows and dampen output although its impact tends to be temporary. In addition, exchange rate depreciation in excess of 1% tends to generate negative effects on the economy.
The rest of the paper is organised as follows. The next section provides a brief overview of the exchange rate policy in Zambia and discusses the evolution of selected macroeconmic and financial indicators. Section 3 reviews relevant literature on macroeconomic effects of exchange rate volatility on the economy. The estimation procedure and empirical results are presented in section 4. Section 5 concludes and offers policy recommendations.
2. Overview of the Exchange Rate Policy and Trends in Selected Macroeconic Indiactors
In recent years, the focus of economic policy has been on raising economic growth and reducing poverty by consolidating the stabilisation gains achieved since the commencement of economic reforms in early 1990s. Financial sector and private sector development plus public financial management and accountability reforms are other areas of priority concern to the Government.
Prior to economic reforms, the focus of Government policy was on industrialisation built around import substitution adopted soon after gaining independence in 1964. The main sector of the economy was mining which accounted for about half of GDP and about 90% of total export earnings (UNDP, 2006)1. Strong copper revenues initially supported the development strategy as copper prices and output steadily rose. Subsequently, revenue from copper weakened due to the fall in copper prices in early 1970s and the decline in copper output. Oil prices shot up in 19973/74 and again in 1979/80 (oil price shocks) causing the terms of trade to decline.
The Government responded by imposing exchange and trade controls (foreign exchange controls on current and capital account transactions plus import licensing) as well as price controls on goods. Interest rate ceilings were also introduced. The exchange rate was fixed while monetary and fiscal policies were loosened. Attempts to diversify the economy away from copper dependence and limit the dominance of state owned enterprises (SOEs) created to spur the industrialisation development strategy in the late 1980s failed due to underlying distortions created by the policies in place (see Adam, 1995). Negative real GDP growth rates became entrenched by the mid 1990s.
Consequently, the economy was liberalised as market-oriented reforms were adopted in early 1990s with the support of the IMF/World Bank. The reform package included the de-control of product prices, trade liberalisation and elimination of exchange controls in 1994 making both current and capital account fully convertible that subsequently gave way to a market determined exchange rate. The removal of interest rate ceilings in 1991, introduction of Treasury bills auctions in 1993 and adoption of indirect instruments of monetary policy in 1995, privatisation of SOEs including mining companies and rationalisation of expenditures
1 Mining continues to be a dominant sector and underpins the performance and structure of the economy (UNDP, 2006).
3
(especially non-essential) and enhancement of domestic revenue collection were other reform measures undertaken.
Zambia has maintained a liberal flexible exchange rate system since 1994. Prior to that, the exchange rate was fixed from the time of independence in 1964. A fixed exchange rate regime existed from 1964 to 1982 and 1987 to 1991 while a crawling peg was adopted between 1983 and 1985. An initial float of the kwacha took place between 1985 and 1987. A more flexible exchange rate regime was adopted in the early 1990s as part of the economic reforms. The decision to choose each of these exchange rate regimes was largely influenced by conventional economic and political arguments. A comprehensive review of the exchange rate policy in Zambia is provided by Chipili (2010), Mungule (2004) and Mkenda (2001).
According to figure 1, the kwacha/US dollar exchange rate exhibited a rising trend over the priod 1994-2013 with some volatility, particularly during the global financial crisis of 2007/8 and the period following debt forgiveness (HIPIC Initiative and MDRI) in early to mid-2000. Over the sample period, remarkable economic improvements have been posted, evidenced by sustained growth in real GDP since early 2000, reflecting strong growth in agricultural output supported by favourable weather and appropriate farmer input support programme by government; recovery in the mining sector following huge re-invetsment after privatising the state owned mining conglomerate; pick-up in tourism due to various promotional activities; and steady growth in the contruction sector owing to the boom in private commercial and housing properties, public investment in health and educational facilities and road infrastucture by the government. In addition, inflation has constitently declined to single digits in the recent past, reflecting the stabilisation programme embarked on by government since the commencement of economic reforms in early 1990. Similarly, commercial bank lending rates have declined in line with the trend in inflation. Money supply growth, though volatile, has slowed down and stabilised around an annual growth rate of 20% in recent years, reflecting the resolve by authorities to contain inflationary pressures while supporting growth.
The current account has recovered since 2000 despite some fluctuations, reflecting strong growth in exports (both copper and non-traditional exports) while imports have been rising in recent years in line with the expansion in key economic sectors such as mining. Despite exhibiting considerable volatility, foreign direct invetsmet has picked up attributed to a stable and robust macroeconomic environment. The size of the capital market has expanded over the years as reflected in the rising stock market index. Fiscal deficits (as a percent of GDP) improved from around 8% in 1994 to around 1% in 2008; however, the trend reversed in recent years reaching around 6% in 2013 largely due to the infrastructure investment programme embarked on by Government. Finally, the declining trend in domestic revenue collections was reversed in 2010, picked up sharply and exceeded the 1994 level by 2013, reflecting steady economic expansion and various revenue broadening measures.
Figure 1 Selected Macroeconomic Indicators
4
0
1
2
3
4
5
6
94 96 98 00 02 04 06 08 10 12
EXCHANGE_RATE_KUS$
-4
-2
0
2
4
6
8
94 96 98 00 02 04 06 08 10 12
REAL_GDP_GROWTH
0
20
40
60
80
94 96 98 00 02 04 06 08 10 12
INFLATION
0
20
40
60
80
94 96 98 00 02 04 06 08 10 12
MONEY_SUPPLY_GROWTH
0
20
40
60
80
94 96 98 00 02 04 06 08 10 12
LENDING_RATE
-20
-16
-12
-8
-4
0
4
8
94 96 98 00 02 04 06 08 10 12
CURR_ACC_PERCENT_GDP
25
30
35
40
45
50
94 96 98 00 02 04 06 08 10 12
EXPORTS_PERCENT_GDP
28
32
36
40
44
48
94 96 98 00 02 04 06 08 10 12
IMPORTS_PERCENT_GDP
0
2
4
6
8
10
12
94 96 98 00 02 04 06 08 10 12
FDI_PERCENT_GDP
0
1,000
2,000
3,000
4,000
5,000
6,000
94 96 98 00 02 04 06 08 10 12
STOCK_MKT_INDEX
14
16
18
20
22
94 96 98 00 02 04 06 08 10 12
REVENUE_PERCENT_GDP
-10
-8
-6
-4
-2
0
94 96 98 00 02 04 06 08 10 12
FISCAL_DEFICIT_PERCENT_G
Note: Exchange_Rate_KUS$=absolute values of kwacha/US$; Real_GDP_Growth=%; Inflation=%;Money_Supply_Growth=%; Lending_Rate=%; Curr_Acc_Percent_GDP=current account in % of GDP;Exports_Percent_GDP=exports as % of GDP; Imports_Percent_GDP=imports as % of GDP; FDI_Percent_GDP=foreign direct investment as % of GDP; Stock_Mkt_Index= stock market index (=1000); Revenue_Percent_GDP=revenue as % of GDP; Fiscal_Deficit_Percent_GDP=fiscal deficit as % of GDP.
3. A Brief Look at the Literature
5
There is no consensus in the literature regarding the factors that influence exchange rate volatility due to divergent theoretical models of exchange rate determination. In addition, the factors are numerous compounded by the inability to predict them and their effect on real exchange rate volatility varies considerably to be quantified with certainty (Canales-Kriljenko and Habermeier, 2004; Dungey, 1999; Korteweg, 1980).
A large body of empirical work has employed fundamentals-based models in modelling exchange rate volatility (see Kočenda and Valachy, 2006; Devereux and Lane, 2003; Bayoumi and Eichengreen, 1998; Pozo, 1992). Most of these studies document real shocks with large permanent effects as the dominant source of exchange rate volatility. Notwithstanding this evidence, fundamentals tend to explain only a small proportion of exchange rate volatility due to the weak link between the two (popularly referred to as the ‘disconnect argument’), thus lending support for the role of microstructure factors in exchange rate fluctuations in the short-term (see Bjørnland, 2008; De Grauwe et al. 2005; Obstfeld and Rogoff, 2000; Flood and Rose, 1995 for an extensive discussion). However, Morana (2009) provides more recent support for fundamentals in terms of linkages (causes and trade-off) between exchange rate volatility and volatility in macroeconomic factors (i.e. output, inflation, interest rate and money supply). Although the direction of causality is bi-directional, it is however, stronger from macroeconomic factors to exchange rate volatility than vice-versa. Thus, stability in the macroeconomic variables is recommended to reduce exchange rate volatility in sharp contrast to Flood and Rose (1995).
In the case of Zambia, the influence of copper price - major commodity export - on the exchange rate like for most developing countries whose currencies are commodity-based is documented (Cashin et al. 2002). In addition, the positive influences of exchange rate regime, money supply and openness on conditional volatility of the kwacha/US dollar is established (see Chipili, 2010; Bangaké, 2008; Hausmann et al., 2006). Savvides (1990) finds both real and monetary variables to be key determinants of exchange rate variability.
Thus, identifying sources of exchange rate volatility facilitates the design of appropriate policy response. By and large, real (external) sources tend to constrain policy response due to their unpredictability of occurrence and highly differential impact on the exchange rate. This is in contrast to the domestically generated factors such as monetary which authorities have reasonable control over. If the sources are monetary in nature, the policy response is to reduce the trend rate in monetary expansion and/or raise trend rate in real output growth. This entails adopting a predictable monetary policy rule that involves pre-announcement of monetary targets, instruments and their timing.
Empirical evidence regarding the impact of exchange rate volatility on macroeconomic variables is mixed. For instance, Baxter and Stockman (1989) do not find evidence of exchange rate volatility impacting on macroeconomic aggregates under alternative exchange rate regimes. Low exchange rate volatility is associated with greater output growth and lower inflation (Robertson and Symons, 1992). Sapir and Sekkat (1995) find no appreciable impact of exchange rate volatility on trade, investment and growth. On the contrary, Chit (2008) argues that evidence of exchange rate volatility adversely affecting investment, growth and trade exists. Schnabl (2009) finds exchange rate volatility to exert negative effect on growth.
The theoretical prediction of the exchange rate volatility-trade relationship is ambiguous: exchange rate volatility can either stimulate or depress trade (Hooper and Kohlhagen, 1978; Baron, 1976; Clark, 1973; Ethier, 1973). Similarly, empirical evidence is inconclusive (see Ozturk, 2006; Clark et al. 2004; McKenzie, 1999; Côté, 1994). Nonetheless, there is growing and firm evidence that exchange rate volatility imposes significant effects on the volume of trade (see Ozturk, 2006; Clark et al. 2004; UK Treasury,
6
2003; McKenzie, 1999; Côté, 1994; IMF, 1984; Farrell et al., 1983). Exchange rate variability affects international specialisation in production which in turn leads to a reduction in the welfare of people as output declines and consequently income and consumption (Clark, 1973). Volatility in the exchange rate can lead to the reduction in the volume of international trade due to increases in the level of trade riskiness that creates uncertainty about profits. In addition, it causes prices of tradables to rise due to the risk mark-up (risk premium) imposed by sellers in order to protect profits. This tends to affect the competitiveness of exports. In response to fluctuations in the exchange rate, firms shift resources from the risky tradable sector to the less risky non-tradable sector in order to protect their profits. Further, a rise in exchange rate uncertainty increases transaction costs as agents attempt to hedge against exchange rate risk (Schnabl, 2009).
The importance of exchange rate in monetary policy and foreign exchange interventions is established. Most empirical work especially in developed countries is dominated by estimations of the interest rate rule proposed by Taylor (1993) as most central banks typically use the nominal short-term interest rate as the main operating policy instrument (see Eleftheriou et al. 2006; Adam et al. 2005; Clarida et al. 1998). Central banks mostly adjust the nominal short-term interest rate in response to deviations of output and inflation from potential level and target, respectively. In Zambia, exchange rate volatility is found to be an integral part of monetary policy setting and central bank foreign exchange intervention decision (see Chipili, 2013).
4. Estimation Method, Data and Empirical Results
To establish the impact of exchange rate volatility on the economy, we begin by estimating exchange rate volatility using the GARCH model. Thereafter, the structural VAR (SVAR) model that includes exchange rate volatility derived from the GARCH model and selected macroeconomic variables is estimated and impulse responses and variance decomposition results reported. Finally, the threshold for exchange rate volatility deemed disruptive is determined using the non-linear effect apparoch.
4.1 Estimation Method
GARCH Model
Volatility in the kwacha/US dollar exchange rate is modelled using the GARCH approach specified in equations 1 and 2 in order to characterise its time-varying conditional variance path. The GARCH models introduced by Engle (1982) and extended by Bollerslev (1986) take into account the distributional form of the exchange rate: leptokurtic, clustering and leverage behaviour typically present in financial time series data (Brooks, 2001; Bauwens and Sucarrat, 2006). The standard deviation, a widely used measure of exchange rate volatility, is not adopted on account of the fact that it tends to overstate total risk, does not distinguish between predictable and unpredictable elements of a variable and lastly assumes normality distribution of a variable which is not always the case especially for financial variables.
7
Δst=α0+∑i=1
q
α i Δst−i+εt (1)
ε t|Ωt−1 ~ N (0 , σ t2 )
σ t2=ht
ht=α0+∑i=1
p
α i εt−i2 +∑
j=1
q
β j ht− j (2)
where Δst is the logarithm of the percentage change in the nominal kwacha/US dollar
exchange rate, ε t is residuals2 which are used to test for ARCH effects based on the Engle (1982) LM test statistic3. Equation 1, specified in log first difference, represents the
conditional mean and describes how st evolves over time. Equation 2 captures the
conditional volatility of st such that α i≥0 ,∀ i=0,1,2,3, . .. . p ; β i≥0 , ∀ i=0,1,2,3, . .. . q ; and
α 0 is the long-term average value of conditional variance.
The persistence of shocks to conditional variance is captured by the sum of α and β . The closer the sum is to 1 the more persistent the shocks are: shocks exert a permanent effect on volatility if α +β=1
(i.e. a unit root in variance is obtained and this kind of GARCH
model is referred to as Integrated GARCH (IGARCH) model). Convergence of conditional volatility to its long-term value is not achieved when α +β>1
or α +β=1 ; instead
volatility is explosive and tends to infinity as shocks persist forever. However, a stationary
GARCH that ensures model stability or convergence of conditional variance forecast to α 0 as
the prediction horizon increases obtains if α +β<1 holds.
SVAR Model
The SVAR approach is employed to determine the effect of exchange rate volatility on selected macroeconomic indiactors. In the SVAR4 set up, all the variables are treated a priori as endogenous and theoretically motivated restrictions imposed on contemporaneous
2 ε t can be generated either from an autoregressive (AR), autoregressive moving average (ARMA) or standard
regression model.
3 To test for the presence of ARCH effects, squared residuals ε¿
t2
are regressed on a constant and q lags (set by the researcher) of squared residuals and the Engle (1982) LM test statistic under the null H 0 : δ0=0 , δ 1=0 , δ2=0 , .. .. . .. , δ q=0 against H1 :δ 0≠0
or δ1≠0
or δ 2≠0
or …. or δ q≠0
is
conducted. LM=TR2 where LM approximate χ
2 (q ) , T is the number of observations while R2
is
computed from the ε¿
t2
equation.
8
relations among variables are examined such that the marginal effect of a shock to any of the variables in the system and on itself can be traced out over time using impulse response analysis5 in a dynamic interaction form.
Thus, the relationship among variables can be set up in VAR form of order p consisting of a system of equations equal to the number of variables (see Clarida and Gertler, 1996)
AX t=A0+∑i=1
p
A1 X t−1+Bε t (3)
where A is an invertible (nxn ) matrix capturing contemporaneous relations among X t
variables; X t is an (nx1 ) vector of macroeconomic variables; A0
is a vector of constants; A1
toAp is (nxn )
matrix of unknown parameters on lagged values of X t to be estimated; B is
an (nxn ) matrix reflecting direct effects of some ε t
on more than one X t variable; ε t
is an
(nx1 ) vector of uncorrelated structural innovations or shocks corresponding to each element
of X t with covariance matrix E( εt εt
' )=Σε ; t=1,2 ,. . .. .. , T ; and n is the number of variables
in the system. Further manipulation yields
X t=Ψ 0+∑i=1
p
Ψ 1 X t−1+e t (4)
where Ψ 0=A−1 A0 ; Ψ 1=A−1 A1 ,Ψ 2=A−1 A2 , .. . ,Ψ p=A−1 Ap ; and e t=A−1 Bε t
is an (nx 1 )
vector of white noise error term with zero mean and constant variance E( et et' )=Σe . Equation
4 is a reduced form representation of equation 3 as the latter cannot be estimated directly since the structural model cannot be identified. Structural shocks are orthogonal to each other
while the reduced form errors, e t , are not. e t are a linear combination of orthogonalised structural shocks.
Thus, the structure linking the structural shocks and the reduced form residuals takes the form
Aet=Bε t (5)
4 SVAR models treat every variable as endogenous due to the difficulty of finding exogenous variables in macroeconomics (Gottschalk, 2001).5 Impulse response functions are calculated from the estimates of the VAR. They show how current and future values of each variable in the VAR respond to a one-off unit increase in the current value of one of the structural shocks in the VAR holding other shocks constant.
9
To identify A and B and thus generate impulse response functions, at least 2 n2−n(n+1 )/2
additional restrictions are required in addition to n normalisation
restrictions. A non-recursive (structural factorisation) identifying structure is adopted whereby a
priori restrictions are imposed on contemporaneous interactions among X t variables in order
to identify the coefficient matrix A . Thereafter, the dynamic impact of ε t can be traced on
the path of any element in X t (Bjørnland and Halvorsen, 2008)6. Thus, the following identification scheme is employed
is real output, pt is the price level – a measure of inflation, impt is imports, xxt
is exports, , k t is a measure of capital flows and erv t
is a measure of exchange rate volatility.
Output, inflation, imports, exports, capital flows and exchange rate volatility shocks are
denoted as ε ty
,ε tp
, ε timp
,ε txx
, ε tk
and ε terv
, respectively.
Diagonal elements of A are normalised to 1 while zero (zero exclusion restriction)
implies no contemporaneous relationship between X t variables7. The study focuses on identifying the impact of a one off shock to exchange rate
volatility on output, inflation, trade (exports and imports) and short-term capital flows. Output adjusts sluggishly with a lag to financial and monetary variables (Brischetto
and Voss, 1999; and Bjørnland and Halvorsen, 2008). Thus zero exclusion restriction is
imposed on all variables in A corresponding to the yt row. Prices adjust slowly to all
6 Matrix A must be identified in order to derive impulse response functions and variance decomposition (Bjørnland and Halvorsen, 2008).
7 a ij = variable j affects variable i instantaneously.
10
variables except movements in output to which they react contemporaneously (Brischetto and Voss, 1999), hence zero exclusion restriction is imposed on all variables corresponding to the pt
row except y t . Imports are driven by domestic income, inflation and exchange rate risk
while exports are influenced by domestic inflation and exchange rate volatility. It is further assumed that capital flows only respond contemporaneously to exchange rate volatility hence
zero exclusion on all variables in the k t row except for exchange rate volatility. Finally, the
exchange rate depends upon innovations in macroeconomic variables as it reacts almost instantaneously to all information. Hence there is no zero exclusion in the row corresponding
to erv t (Brischetto and Voss, 1999; and Bjørnland and Halvorsen, 2008)8.
To examine the threshold for exchange rate volatility beyond which volatility is deemed to be disruptive on the economy, a non-linear size-dependent effect speficification is employed as follows
Δy t=γ 0+γ1 Δst+γ2 Δst . Zc (7)
where Z={1, if |∆ s|c>00 ,if |∆ s|c<0
such that c is a threshold representing a certain percentage rate of depreciation in the
kwacha/US dollar exchange rate, and Δst is the logarithm of the percentage change in the
nominal kwacha/US dollar exchange rate (an approximate measure of volatility).The motivation here is to determine what happens to y when exchange rate
depreciation exceeds a certain level which defines the threshold of exchange rate volatility beyond which undesirable effects on y are generated. Different values of c are used to determine the threshold level of exchange rate volatility. The threshold value c is the one
that maximises R2
. The choice of c is guided by the average rate of depreciation of the kwacha/US dollar over the sample period, equal to 2.5%. The initial value of c is set at 0.5% and subsequently increased at half a percent until 2.5% after which further values of c are set at 3%, 4% and 5% (which is on the uppe limit of the rate of depreciation).
4.2 Data Sources and Description
Monthly data from 2007 to 2013 are used to estimate the impact of exchange rate volatility on output, inflation, trade (exports and imports) and short-term capital flows due to
8 Allowing fxpt and fxst
to contemporaneously affect erv t is rejected.
11
data limitations on macroeconomic and financial varaibles. All the data are sourced from the Bank of Zambia.
As data on real output, GDP, are not available at monthly frequency, the quadratic-match sum interpolation method is used to derive monthly real GDP ( y ) from the annual series. This method ensures that monthly values sum up to the annual value and allows for some variation in a year as opposed to alternative constant-linear and linear-match last methods. The consumer price index (CPI , 2009=100 )is used as a measure of inflation( p) . Exports ( xx )and imports ( imp )
data enter separately in the estimated relationship as
exchange rate volatility may have differential impact on them. Short-term capital flows (k ) is measured by non-resident holdings of fixed income securities issued by the government ( govhold ) . Foreign direct investment data are not available at monthly interval hence the use of govhold . Exchange rate volatility (erv ) of the kwacha/US dollar is derived from the GARCH model.
4.3 Empirical Results
The empirical results from the GARCH model reported in table 1 and plotted in figure 2 reveal that the kwacha/US dollar exchange rate is characterised by episodes of high volatility over the sample period. In particular, volatility was high soon after the foreign exchange market was liberalised in 1994, a reflection of exchange rate adjustment following a long period of misalignment. Volatility baceme more pronounced in 2001 and for most part in 2003. Therefater, volatility receded until the global financial crisis in 2007 and persisited until the end of 2010. A spike in volatility occurred towards the end of 2012. Over the same period, GDP, exports and imports rose while inflation slowed down from double to single digits. Non-resident holdings of fixed income securities varied considerably.
Table 1 GARCH Model ResultsΔst = 0.503 + 0.442Δst−1
(2.54)** (6.47)*** ht = 2.218 + 0.323ε t−1
2
+ 0.516ht−1 (5.27)*** (3.79)*** (7.36)***
Diagnostic testsυt Q(5 )=3.1992[0.669];
υt2Q(5)=0.9093[0.970]; J-B =108.0861[0.000]; ARCH LM
=0.0723[0.7883]; Log L =-607.221; AIC =-5.145; SBC =5.218
The lag length for υt Q and υt2 Q
of 5 is determined according to Tsay (2002): k=ln (T )
where k is lag length and T is the number of observations. z-statistics are reported in parenthesis while p-values are in square brackets. ***,**,* refer to statistical significance at 1%, 5% and 10%, respectively.
12
Note: The GARCH model is well specified based on the diagnostic tests. Volatility in the kwacha/US dollar exchange rate is also found to be persistent.
Figure 2 Plots of Kwacha/US dollar Exchange Rate Volatility and Selected Macroeconomic Variables
13
Unit root tests precede the estimation of SVAR models. The unit root tests reported in table 2 are conducted using ADF and P-P methods. All the variables are I(1) except exchange rate volatility. Nonetheless, the SVAR model is estimated as its implementation requires at least one variable to be I(1) (see Enders, 2004)9. All the variables enter the SVAR in log levels except erv to allow for possible cointegration among them (Lewis, 1995). A lag of one is used in the VAR based on AIC and SBC.
9 SVAR results are sensitive to underlying identifying assumptions, variations to sample length and lags in the VAR.
14
xx -2.21 -20.59*** 1 C&T -2.26 -20.89*** C&T
imp -2.61 -15.10*** 1 C&T -3.79 -30.66*** C&T
govhold -2.26 -5.14*** 1 C&T -1.90 -5.10*** C&T
Critical values for unit root tests are Mackinnon (1996) one-sided p-values. All variables are expressed in natural logarithm except exchange rate volatility (erv ) and non-resident purchase of equities at the stock exchange (stock ). *, ** and *** imply 1%, 5% and 10% levels of significance, respectively. C is constant while T stands for trend.
As shown in table 3, diagnostic tests from the estimated SVAR model indicate the absence of serial correlation of order two. The over-identification restrictions under the null hypothesis that the restrictions are valid with a test statistic that has a chi-square distribution with three degrees of freedom cannot be rejected at 5% significance level.
Table 3 SVAR Model Results: Estimated A Matrixy p imp xx govhold erv
1 0 0 0 0 0-0.13
(-0.37)1 0 0 0 0
-77.77(-0.04))
13.87(0.05)
1 0 0 -0.14(-0.03)
0 -3.97(-1.15)
0 1 0 0.01(0.26)
0 0 0 0 1 -0.001(-0.46)
11108.39
(0.01)
-9035.72(-0.01
-1516.24(-0.01)
509.16(0.01)
16.89(0.01)
1
Test for over-identification restrictions: χ2 (3)=6 .361[ 0 .0953 ] Sample period: 2007.10-2013.12. z-statistics are in parenthesis. VAR (1) diagnostics: serial correlation LM test
of order 2=30.179[0.7412]; ARCH χ2
=723.9529[0.0000]; J-B normality test=1929.598[0.0000] with df=12. Lag length 1 for the VAR was chosen on the basis of AIC and SBC.
The impulse response function results of one standard deviation innovation in exchange rate volatility over a five -year horizon are presented in figure 3. The results suggest that shocks to the kwacha/US dollar exchange rate volatility have an immediate negative effect on exports, output and inflation but imports respond with a lag. Non-residents response countercyclically initially to the exchange rate volatility shock by increasing their holdings of fixed income securities; however, the portfolio holdings gradually reduce as the exchange rate risk is assimilated in investment portfolio decision. Overall, the effect on exchange rate volatility shock on the variables understudy tends to be temporary. As exports recover from a shock to exchange rate volatility, output too improves and gradually returns to steady-state. Imports fall some months after the shock to exchange rate volatility has occurred, but later recover and return to steady-state. Inflation increases on impact in response to unexpected increase in exchange rate volatility which generates higher uncertainty and subsequently affects inflation expectations. Inflation continues to rise, peaking after about
15
six months and then gradually tends towards its equilibrium, reflecting that short-lived shocks are not incorporated in underlying inflation (Taylor, 1981).
Figure 3 Impulse Response Analysis
The proportion of the error of forecast for 1, 6, 12, 24, 36 and 60 months forecast horizon attributed to shocks to exchange rate volatility is reported in table 4. This determines the fraction of the variation in output, inflation, imports, exports and capital flows due to
16
exchange rate volatility shocks. The results reveal that shocks to exchange rate volatility have a stronger influence on trade (both at short and long horizons) followed by inflation, output and lastly capital flows. This underscores the importance of the exchange rate in trade and monetary policies and the transimission of exchange rate impulses to the rest of the economy.
Table 4 Variance DecompositionVariance decomposition of selected
macrovariables in response to exchange rate volatility shock
A summary of the size-dependent results is presented in table 5 while detailed estimations are reported in the appendix. The results in table 5 reveal that, by and large, persistent depreciation in exchange rates in excess of 1% per month have a tendency to generate undesirable effects on the economy. Exchange rate depreciation in excess of 1% will have a depressing effect on exports which in turn impacts on output at the same rate of exchange rate depreciation of 1%. It is however, noted that import demand is less sensitive to exchange rate depreciation than exports, responding strongly when exchange rate depreciation exceeds 4% consistent with the SVAR model result. Thus, with a high rate of depreciation of the kwacha, import prices expressed in domestic currency rise and consequently lead to higher domestic inflation. The inelasticity of imports demand could be attributed to uncertainity regarding persistence in exchange rate depreciation: agents are unlikely to change their behavior if they believe that the currency depreciating trend is transistory. The high sensistivity of exports to exchange rate depeciation could reflect the risk averseness of exporters seeking to protect profits. Finally, short-term capital flows tend to respond negatively to exchange rate volatility once exchange rate depreciation exceeds 1%.
securities Exports Output Imports Inflation0.5%1.0% x x x1.5%2.0%2.5%3.0%4.0% x x5.0%
5. Conclusion The study analysed the impact of exchange rate volatility on output, inflation, trade
and short-term capital flows in Zambia over the period 2007-2013. The extent of exchange rate volatility deemed disruptive on the economy was also determined.
The results reveal that volatility in the kwacha/US dollar exchange rate tends to depress trade by reducing both total exports and imports and subsequently impact on output and inflation. Short-term capital flows are also discouraged by exchange rate fluctuations. Further, exchange rate depreciation in excess of 1% per month tends to generate negative effects on the economy.
The results from this study underscore the importance of the exchange rate in trade and monetary policies and the transimission of exchange rate impulses to the rest of the economy. Exchange rate volatility forms an essential part of trade, monetary and exchange rate policy formulation and implementation and in attracting portfolio investments to ensure that a stable macroeconomic environment is achieved and maintained.
Thus, the key lesson from this study is that central banks must stabilize exchange rates; however, the choice of the currency stabilization strategy matters. Attempts by the central bank to stabilize the exchange rate or avoid sharp currency movements through the direct use of official reserves to intervene in the market may lead to significant deterioration of reserves which in turn erode the country’s ability to respond to adverse external shocks. Thus, complementary policies are recommended to achieve and maintain a stable exchange rate: fiscal policy that ensures sustainable expenditure; monetary policy to remain focused on inflation control of which exchange rate is a relevant factor in inflation dynamics; and a general stable macoeconomic, political and social environment as the exchange rate is influenced by expectations especially in the short-term in response to economic, political and social news.
References
18
Adam, C. (1995) ‘Financial Liberalisation and Currency Demand in Zambia’, Journal of African Economies, 8(3): 268-306.
Adam, C., D. Cobham and E. Girardin (2005) ‘Monetary Frameworks and Institutional Constraints: UK Monetary Policy Reaction Functions, 1985-2003’, Oxford Bulletin of Economics and Statistics, 67(4): 497-516.
Aghion, P., P. Bacchetta, R. Ranciere and K. Rogoff (2009) ‘Exchange Rate Volatility and Productivity Growth: The Role of Financial Development’, Journal of Monetary Economics, 56: 494-513.
Arrtibel, O., D. Furcen, R. Martin, and A. Zdzienicka (2011) ‘The Effect of Nominal Exchange Rate Volatility on Real Macroeconomic Performance in the CEE Countries’, Economic Systems, 35: 261-277.
Bangaké, C. (2008) ‘Exchange Rate Volatility and Optimum Currency Area: Evidence from Africa’, Economics Bulletin, 6(12): 1-10.
Bauwens, L., D. Rime and G. Sucarrat (2006) ‘Exchange Rate Volatility and the Mixture of Distribution Hypothesis’, Empirical Economics, 30: 889-911.
Baxter, M. and A. Stockman (1989) ‘Business Cycles and the Exchange Rate Regime. Some International Evidence’, Journal of Monetary Economics, 23: 377-400.
Bayoumi, T. and B. Eichengreen (1998) ‘Exchange Rate Volatility and Intervention: Implications of the Theory of Optimum Currency Areas’, Journal of International Economics, 45: 191-209.
Bergin, P. (2004) ‘Measuring the Costs of Exchange Rate Volatility’, FRBSF Economic Letter, 22, August.
Bollen, B. (2008) ‘Long-Term Asymmetry in the USD-DEM Spot Exchange Rate Volatility Process’, Applied Financial Economics Letters, 4: 403-407.
Bollerslev, T. (1986) ‘Generalized Autoregressive Conditional Heteroskedasticity’, Journal of Econometrics, 31: 307-327.
Brischetto, A. and G. Voss (1999) ‘A Structural Vector Autoregression Model of Monetary Policy in Australia’, Reserve Bank of Australia Research Discussion Paper No.1999-11.
Brooks, C. (2006) Introductory Econometrics for Finance, Cambridge: Cambridge University Press.
Canales-Kriljenko, J. and K. Habermeier (2004) ‘Structural Factors Affecting Exchange Rate Volatility: A Cross-Section Study’, International Monetary Fund Working Paper No.147.
Cashin, P., L. Céspedes, R. Sahay (2002) ‘Keynes, Cocoa, and Copper: In search of Commodity Currencies’, IMF Working Paper WP/02/223.
Chipili, J. M. (2010) ‘Macroeconomic Effects of Exchange Rate Volatility in Zambia’, PhD Thesis, University of Leicester.
Chipili, J. M. (2013) ‘Monetary Policy, Foreign Exchange Intervention and Exchange Rate Volatility in Zambia’, The African Finance Journal, 15(1): 36-55.
Chipili, J. M. (2013) ‘Exchange Rate Volatility and Trade Flows in Zambia’, African Development Review, 25(1): 55-66.
Chit, M. M. (2008) ‘Exchange Rate Volatility and Exports: Evidence from the ASEAN-China Free Trade Are’, Journal of Chinese Economic and Business Studies, 6(3): 261-277.
Clarida, R. and M. Gertler (1996) ‘How the Bundesbank Conducts Monetary Policy’, Economic Research Reports C.V. STARR Center for applied Economics, RR No.96-14, April.
Clarida, R., J. Gali and M. Gertler (1998) ‘Monetary Policy Rules in Practice: Some International Evidence’, European Economic Review, 42: 1033-1067.
19
Clark, P. B. (1973) ‘Uncertainty, Exchange Risk and the Level of International Trade’, Western Economic Journal, 11: 302-313.
Clark, P. N., Tamirisa, Shang-Wei, A. Sadikov and L. Zeng (2004), ‘Exchange Rate Volatility and Trade Flows: Some New Evidence’, IMF, May.
Côté, A. (1994) ‘Exchange Rate Volatility and Trade – A Survey’, Bank of Canada, Working Paper 94-5, May.
Crosby, M. (2000) ‘Exchange Rate Volatility and Macroeconomic Performance in Hong Kong’, Department of Economics, University of Melbourne and Hong Kong Institute for Monetary Research.
De Grauwe, P., D. Roberto and G. Marianna (2005) ‘Fundamental and Non-Fundamental Equilibria in the Foreign Exchange Market. A Behavioral Finance Framework’, CESifo Working Paper No.1431.
Devereux, M. B. and P. R. Lane (2003) ‘Understanding Bilateral Exchange Rate Volatility’, Journal of International Economics, 60: 109-132.
Dickinson, D. and J. Liu (2007) ‘The Real Effects of Monetary Policy in China: An Empirical Analysis’, China Economic Review, 18: 87-111.
Dungey, M. H. (1999) ‘Decomposing Exchange Rate Volatility around the Pacific Rim’, Journal of Asian Economics, 10: 525-535.
Eleftheriou, M., D. Gerdesmeier and B. Roffia (2006) ‘Monetary Policy Rules in the Pre-EMU Era, Is there a Common Rule?’ European Central Bank Working Paper Series No.659, July.
Enders, W. (2004) Applied Econometric Time Series, New York: Wiley.Engle, R. F. (1982) ‘Autoregressive Conditional Heteroskedasticity with Estimates of
the Variance of the United Kingdom Inflation’, Econometrica, 50: 987-1007.Ethier, W. (1973) ‘International Trade and the Forward Exchange Market’, American
Economic Review, 63: 494-503.Farrell, V. S., D. A. DeRosa and T. A. McCown (1983) ‘Effects of Exchange Rate
Variability on International Trade and Other Economic variables: A Review of the Literature’, Board of Governors of the Federal Reserve, Wahington, D.C., December.
Flood, P. R. and A. Rose (1995) ‘Fixing Exchange Rates: A Virtual Quest for Fundamentals’, Journal of Monetary Economics, 36: 3-37.
Gottschalk, J. (2001) ‘An Introduction into the SVAR Methodology: Identification, Interpretation and Limitations of SVAR Models’, Kiel Institute of World Economics, Kiel Working Paper No.1072.
Hausmann, R., U. Panizza and R. Rigobon (2006) ‘The Long-Run Volatility Puzzle of the Real Exchange Rate’, Journal of International Money and Finance, 25: 93-124.
Hooper, P. and S. Kohlhagen (1978) ‘The Effect of Exchange Rate Uncertainty on the Price and Volume of International Trade’, Journal of International Economics, 8: 483-511.
IMF (1984) ‘Exchange Rate Volatility and World Trade’, International Monetary Fund Occasional Paper No.28.
Kočenda, E. and J. Valachy (2006) ‘Exchange Rate Volatility and Regime Change: A Visegrad Comparison’, Journal of Comparative Economics, 34: 727-753.
Korteweg, P. (1980) ‘Exchange Rate Policy, Monetary Policy and Real Exchange-Rate Variability’, Essays in International Finance, No.140, Pricenton.
Lewis, K. K. (1995) ‘Are Foreign Exchange Intervention and Monetary Policy Related, and Does it Really Matter?’ The Journal of Business, 68(2): 185-214.
McKenzie, M. D. (1999) ‘The Impact of Exchange Rate Volatility on International Trade Flows’, Journal of Economic Surveys, 13(1): 71-106.
Mirdala, R. (2012) ‘Macroeconmic aspects of Real Exchange Rate Volatility in the Central European Countries’, Journal of Applied Economic Sciences, 2(20): 163-178.
20
Mkenda, B. K. (2001) ‘Long-Run and Short-Run Determinants of the Real Exchange Rate in Zambia’, Working Papers in Economics No.40, Department of Economics, Goteborg University.
Morana, C. (2009) ‘On the Macroeconomic Causes of Exchange Rate Volatility’, International Journal of Forecasting, 25: 328-350.
Mungule, K. O. (2004) ‘The Determinants of the Real Exchange Rate in Zambia’, AERC Research Paper No.146.
Musonda, A. (2008) ‘Exchange Rate Volatility and Non-traditional Export Performance: Zambia, 1965-1999’, AERC Research Paper 185.
Obstfeld, M. and K. Rogoff (1998) ‘Risk and Exchange Rates’, NBER Working Paper No. 6694.
Obstfeld, M. and K. Rogoff (2000) ‘New Directions in Open Macroeconomics’, Journal of International Economics, 50: 117-153.
Ozturk, I. (2006) ‘Exchange Rate Volatility and Trade: A Literature Survey’, International Journal of Applied Econometrics and Quantitative Studies, 3(1): 85-102.
Robertson, D. and J. Symons (1992) ‘Output, Inflation and the ERM’ Oxford Economic Papers, 44(3): 373-386.
Sapir, A. and K. Sekkat (1995) ‘The Impact of Exchange Rate Fluctuations on European Community Trade’ available at: http://ideas.repec.org/p/bon/bonsfb/282.html : accessed on 5 May 2009.
Savvides, A. (1990) ‘Real Exchange Rate Variability and the Choice of Exchange Rate Regime by Developing Countries’, Journal of International Money and Finance, 9: 440-454.
Schnabl, G. (2009) ‘Exchange Rate Volatility and Growth in Emerging Europe and East Asia’, Open Economic Review, 20: 565-587.
Sengupta, J. T. (2002) ‘Modelling Exchange Rate Volatility’, University of California, Santa Barbara, Department of Economics, Departmental Working Paper 12-96.
Takagi, S. And Z. Shi (2011) ‘Exchange Rate Movements and Foreign Direct Investment (FDI): Japanese Investment in Asia, 1987-2008’, Japan and The World Economy, 23: 265-272
Taylor, J. B. (1981) ‘Stabilisation, Accommodation, and Monetary Rules’, The American Economic Review, 71(2): 145-149.
Taylor, J. B. (1993) ‘Discretion Versus Policy Rules in Practice’, Carnegie-Rochester Conference Series on Public Policy, 39: 195-214.
Tenreyro, S. (2007) ‘On the Trade Impact of Nominal Exchange Rate Volatility’, Journal of Development Economics, 82(2): 485-508.
Tsay, R. S. (2002) Analysis of Financial Time Series, New York: John Wiley & Sons.UNDP (2006) ‘Economic Policies for Growth, Employment and Poverty Reduction.
Case Study of Zambia’, Lusaka, available at: http://www.undp.org.zm: accessed on 10 June 2008.
Willett, T. D. (1982) ‘The Causes and Effects of Exchange Rate Volatility’, in The International Monetary System, pp24-64, edited by Dreyer, J., G. Haberler and T. D. Willett, Washington D.C., American Enterprise Institute.
