Upload
qmul
View
0
Download
0
Embed Size (px)
Citation preview
1
Exchange rate shocks and monetary policy in Russia
Brigitte Granville and Sushanta Mallick Centre for Globalisation Research
School of Business and Management Queen Mary, University of London
Mile End Road, London E1 4NS, UK
Abstract
The failure of the Russian central bank to control inflation is striking. Russian
monetary policy has failed persistently to achieve sustained low inflation, both in
absolute terms and relative to the peer group of countries similarly exiting from
Soviet-style central planning. The question raised in this paper is why? What kind of
monetary policy has been followed by the central bank during the period 1995 to
2007? Our contribution is to search for a possible transmission channel between the
real interest rate, inflation rate, exchange rate, output growth and foreign reserve
growth, after having controlled for the effect of oil price inflation. Using a vector
autoregressive model in error-correction form and using sign restrictions, we show
that the monetary authorities’ failure to abate double-digit inflation appears to be
driven by the policy of exchange rate targeting, as reflected in our identified exchange
rate shocks.
I. Introduction
In recent years, for most central banks around the world, monetary policy – whether
expressed in terms of interest rates or growth of monetary aggregates – has been
increasingly geared towards the achievement of price stability and low inflation. Very
high or hyperinflation seems to have disappeared, at least since the mid 1990s.
2
Average inflation in developing countries (including many countries of Eastern
Europe and Latin America with histories of high inflation) has declined from triple-
digit figures in the late 1980s to low single-digit figures by the end of 2001 (Domaç
and Yücel, 2005). Wynne and Kersting (2007: 2) report that inflation in the last three
decades has dropped dramatically, averaging just 5.8 percent in developing countries
and 2 percent in industrial countries since 2000. It is in this context of relatively low
global inflation that the double-digit Russian inflation stands out. In 2007, the
consumer price index increased by 11.9%, which is significantly higher than 9.0% in
2006. The government’s inflation forecast at the beginning of that year (2007) was
8.0%. In a global environment where inflation outcomes have dramatically improved
independent of the monetary framework for achieving price stability – that is, both in
countries that have adopted inflation targets as well as in countries that have not (Ball
and Sheridan, 2005) – the failure of the Russian central bank to control inflation is
striking.
This paper aims to identify the true strategy pursued by the Bank of Russia:
targets are announced for both the exchange and the inflation rates (respectively, a
managed float (Keller and Richardson, 2003) and a twelve-month CPI target)
reflecting therefore a problematic mix (Vdovichenko and Voronina, 2006). These
targets seem to be often missed with no consequence for the monetary authority.
Stabilizing the exchange rate has become a major monetary policy goal in a number
of CIS countries, namely Belarus, Kazakhstan, Russia and Ukraine, (Bauer and Herz,
2007).
In the absence of well-developed financial markets (Kutan and Brada, 2000),
capital inflows causes serious problems for domestic monetary policies. With rising
capital inflows, the central bank is more likely to buy excess foreign exchange supply
3
on the market, leading to monetary expansion and inflation. On the other hand, if the
central bank does not intervene in the foreign exchange market, the result could be
nominal appreciation of the local currency which may lead to deterioration of the
current account combined with reduced profitability. At the same time, repatriated
proceeds from the export of oil and gas are quite successfully sterilized by means of
oil price-related variations in the marginal rates of taxes specific to the oil sector, the
proceeds of which flow into the stabilization fund.
Against this background, we search for a possible relation between the real
interest rate, inflation rate, exchange rate, reserve growth, output growth (proxied by
growth in industrial production) and oil prices to indicate what type of monetary
policy has been followed by the Bank of Russia during the period 1995 to 2007. We
test whether the real interest rate is reacting to policy driven exchange rate shocks. If
this is the case, then it can be demonstrated that the monetary policy stance of the
central bank has not been primarily designed for price stability. Our focus is informed
by the argument that inflation persistence is shown to come from economic agents’
limited information about the central bank’s policy objectives (Erceg and Levin,
2003). Thus in this paper, by imposing sign restrictions on impulse responses, we
identify exchange rate shocks and then examine the impulse response functions.
The structure of this paper is as follows. Section 2 gives a snapshot of the story
of the Russian inflation and derives an inflation equation. A monetary policy rule in
an open economy is formulated in order to test empirically the reaction of the real
interest rate to exchange rate changes, inflation, and output and reserve growth.
Section 3 then presents the data, methodology and empirical results. Using both a
vector error correction (VEC) and Uhlig (2005)’s ‘pure sign restriction’ modelling
strategy, this paper suggests that the authorities have targeted the exchange rate –
4
specifically, accumulating foreign reserves in an attempt to prevent nominal
appreciation, but the net effect of what might be termed this weak rouble policy has
become a rise in inflation. Section 4 concludes.
II. Modelling the inflation Dynamics
We have written extensively on the repeated episodes of high inflation in post-
communist Russia (Granville (1995), Ferguson and Granville (2000), Granville
(2001), Granville and Mallick (2006), explaining the constraints – namely the budget
deficit (until 2000 when the fiscal adjustment was made) and the exchange rate –
faced by the central bank in its conduct of monetary policy. While the literature is
extensive, naming a few include Choudhry (1998), Cotarelli and Doyle (1999)
Esanov, Merkl and de Souza (2005), Kutan and Brada (2005), Oomes and Ohnsorge
(2005), Papazoglou and Pentecost (2004), Starr (2005) and Vdovivhenko and
Voronina (2006), our aim in this section is not to offer a review of this literature but
rather a brief summary of our work aimed primarily at explaining why we see the
Russian authorities’ preoccupation with the exchange rate as hampering monetary
policy in its task of achieving low inflation.
A. A Snapshot of the Inflation Dynamics
In the early period 1992-93, an effective monetary policy framework was lacking due
not only to the challenge of establishing new institutions and regulations, but
especially also to the difficulty of overcoming the legacy of central planning where
budget and credit financing were indistinguishable. The absence of domestic financial
markets (the T-bills market was introduced in May 1993) and access to international
5
capital meant that the government budget deficits (consistently over 5% of GDP year
after year) were financed borrowing from the central bank. This ‘monetary financing’
led to very high inflation. The average monthly inflation rate reached 41 percent in
1992 due to the price jump of over 200 percent in January 1992 following price
liberalization (that average monthly figure would be 18 percent without the January
1992 observation), and 21 percent in 1993 (Granville, 2001:100). In our empirical
analysis, we decided to omit 1992 and 1993 in the hope that by omitting these first
years of transition, the data will produce a less noisy measure to extract the results of
the central bank’s performance.1
The hard inflationary times of this early period eventually forced the
government into a stabilization effort, culminating in the signing of a fully-fledged
IMF programme on 26 March 1995, and in April that year, the independence of the
central bank was increased allowing for the institutional separation of monetary and
fiscal policy.2 The central bank stopped officially acting as the banker of the
government and state firms.3
It is during this period that the exchange rate started to act as the other major
constraint for monetary policy – especially after T-bill issuance replaced seigniorage
as the deficit financing mechanism. From early 1995, there was a shift in the
financing of the budget deficit away from central bank credits in favour of short-term
1 Although our sample starts from 1995, in calculating growth rates of different variables, 1994
observations have been used as the initial values.
2 The “Central Bank Law” enacted in April 1995.
3 http://www.cbr.ru/eng/today/history/central_bank.asp: “On April 26, 1995, the Bank of Russia
stopped extending loans to finance the federal budget deficit and centralised loans to individual
industries and sectors of the economy.”
6
treasury bills (or ‘GKOs’).4 The curbing in the growth of domestic credits brought a
sharp appreciation of the domestic currency – the rouble – in April-May (Granville,
2001:108). This led to the adoption in July 1995 of an exchange rate target as the
nominal anchor for the price stabilisation programme. This stabilisation drive resulted
in lower average monthly inflation compared to the previous sub-period (1.85%
compared with 13.18%). But less than three years later, in August 1998, Russia was
thrown back into a profound financial crisis. Why did this happen?
The answer is that substituting bond financing for monetary financing of
stubbornly high budget deficits was like papering over the cracks. Just as Sargent and
Wallace (1981) said would happen, it was ultimately doomed, with the catalyst of
doom proving to be the exchange rate target. In Basdevant and Hall (2002) exchange
rate expectations are shown to have played a key role in the origin of the August 1998
financial crisis.
Between May 1995 and July 1998, the central bank was demonstrating its
commitment to holding the exchange rate band by raising the refinance rate to
whatever level necessary for attracting capital inflows. The refinance rate acted as an
effective cap on the T-bill yield and so signalled the level at which the central bank
would support the price. The calculation was that capital inflows would be attracted
by the real returns available from the combination of high nominal yields on rouble-
denominated debt and the promise of a stable exchange rate. Unfortunately, the Asian
financial crisis eroded risk appetite, and investors in government papers (T-bills) were
deterred by higher interest rates in face of low or zero growth prospects, which
created doubts that the government could afford debt service costs reaching 40 per
cent of federal expenditures in May 1998. The ex-post real interest rate reached
4 1995 Federal Budget Law.
7
unsustainable levels as in May 1998 when the refinance rate was raised from 42
percent to 150 percent.
In this environment, all demand, domestic and foreign, for new issues of rouble
T-bills disappeared. The government could no longer pay debt with debt. Redemption
of weekly maturities averaging about Rbs 9 billion ($1 billion) had to be financed out
of general taxation. The problem was exacerbated by a fall in oil prices and a rise in
imports which led to the first current-account deficit since 1993, while putting further
pressure on fiscal revenues. The IMF-led rescue package agreed in mid-July 1998,
although worth some $22.6 billion, failed to reassure investors and therefore to bring
interest rates low enough to achieve a major expenditure reduction. The only possible
rescue now was drastic cuts in non-debt-service expenditure – that is, ordinary public
spending. The Finance Minister at the time, Mikhail Zadornov, warned legislators that
the pain of making such cuts would be nothing compared to the pain which would
follow a financial collapse. Zadornov’s warning went unheeded, and a sovereign
default duly occurred on 17 August 1998, soon followed by the abandonment of the
exchange rate peg (Granville, 2001: 120). That triggered a chaotic 60% devaluation,
in turn producing a resurgence of inflation.
The 1998 crisis proved a positive turning point leading by 2000 to a federal
budget surplus. The fiscal situation improved due to a radical change in
macroeconomic policy, forced – ironically enough – on the new communist-backed
government by the loss of credit-worthiness after the default. Fiscal surpluses have
replaced fiscal deficits removing one constraint for monetary policy. Fiscal surpluses
have been achieved thanks to a combination of fiscal discipline (in a radical departure
from previous practice) and high oil prices – which, at the same time, have generated
serious problems for monetary and exchange rate policy. Faced with a balance of
8
payments surplus, the monetary authorities have been pulled between the goal of
reducing inflation and the goal of restraining the real appreciation of the rouble.
Starting in 2000, the control of inflation took second place to a policy expedient
deemed necessary to protect jobs and promote output. This expedient was an
exchange rate target. The stability of the real exchange rate is one of the objectives
stated by the central bank in its monetary programme.5 This stated goal appears to be
aimed mainly at protecting output and jobs. Vdovichenko and Voronina (2006),
estimating a monetary rule for the period 1999 to 2003, show that slowing the real
exchange rate appreciation to shelter domestic producers and employment from
import competition has been given priority over reducing inflation.
During the period 2000-2007, Russia achieved annual real GDP growth rates of
around 7 percent, current account surpluses of 10 percent, and fiscal surpluses of 4
percent on average. In an environment of oil-driven balance of payments surpluses,
exchange rate targeting meant artificially restraining the natural appreciation of the
rouble to preserve competitiveness – not only in the most obvious area in this context,
which is the natural resource exporting sectors which account for the bulk of Russian
GDP and over half of the revenues of the federal budget, but also and perhaps more
especially in the import-substituting industries which are major employers. This
‘weak rouble’ policy has been pursued in the form of massive rouble interventions in
the foreign exchange market. The result has been higher inflation.
B. The Inflation Equation6
5 http://www.cbr.ru/eng/today/publications_reports/on_2005e.pdf: 3.
6 Variable definitions are reported in Appendix 1.
9
We consider price formation in the context of an open economy, therefore the
exchange rate plays an important role in both demand and supply disturbances.7 The
improvement in Russia’s external liquidity position has been spectacular in the recent
years. The substantial reserves accumulation reflected a combination of significant
trade surpluses and foreign investment inflows. That also resulted in the relative
strength of the domestic demand, and money supply growth has been much higher
than nominal GDP. Thus we need to derive a reaction function by the monetary
authority to changes in international reserves. The first relationship is the flow supply
of money ( s
tM ) that comes from the central bank’s balance sheet (where tR and
tDA are respectively international reserves and domestic assets) as given by equation:
[1] s
t t tM R + DA∆ = ∆ ∆
The second relationship defines the flow demand for money. The level of real output
(y), real interest rate (r) and exchange rate (e) determine the money demand (Md)
growth. The money demand can thus be shown via an open economy LM function as
follows:
[2] t
d
t t t tM y r eθ β δ ε∆ = ∆ − − ∆ +
where δβθ ,, are all positive. The real interest rate equals )( ei π− , where πe is
expected inflation and i is nominal interest rate. And εt is a demand disturbance. As
real interest rate is a linear combination of the two variables, it can be stationary in
line with the other first-differenced variables and hence it can be left in levels. If we
were to link equation (2) to the Cagan model of money demand under hyperinflation,
7 Starr (2005: 446) also assumes the economies of the CIS to be open.
10
then the parameter β will be of interest, as it would capture the effect of the expected
inflation rate.8
The third key relationship in this model is the assumption of money market flow
equilibrium, which continuously holds:
[3] d sM M∆ = ∆
The capital flow reaction function can thus be derived as:
[4] t t t t t tDA y r e Rθ β δ ε∆ = ∆ − − ∆ − ∆ +
In equilibrium, equation 4 shows that there is an inverse relationship between
changes in international reserves and domestic assets under a fixed exchange rate
regime, which means the central bank has to sterilize the capital inflows to keep the
money supply constant. But, with a flexible exchange rate regime, currency
fluctuations are likely to influence the conduct of monetary policy. Domestic residents
may also hold foreign currency for transactions or precautionary purposes in the
presence of domestic inflation meaning that domestic currency depreciation may lead
to a decline in real money balances encouraging currency substitution (Papazoglou
and Pentecost, 2004). Money demand in dollarized economies often appears to be
highly unstable, making it difficult to forecast and control inflation (Oomes and
Ohnsorge, 2005). This situation where real money balances are influenced by
expected inflation is partly in line with a Cagan style relation under conditions of
hyper inflation (see for example Taylor (1991) and Frankel and Taylor (1993) who
also include currency depreciation in the estimation of the money demand function for
high inflation countries). Choudhry (1998) found that the rate of change of the
exchange rate needs to be included in the demand function for M2 to obtain a
8 See Taylor, 1991; Phylaktis and Taylor, 1993, for more details on Cagan model.
11
stationary long-run relationship. Thus under a flexible exchange rate system, we need
to assess the relative impact of changes in foreign assets on the monetary side in
response to exchange rate shocks. In fact, as the economy grows, there could be
secular growth in monetary expansion.
The aggregate supply equation can be formulated following an open-economy
Phillips curve:
[5] O
t t t t t ty e DAπ λ φ ω ηπ ν= + ∆ + ∆ + +
where the coefficients λ and φ are greater than zero. υt is a supply disturbance. πO is
oil price inflation. The inclusion of DA in this equation suggests that if the exchange
rate does not change in response to changes in capital flows, higher money growth
can contribute to generating inflation. The changes in oil prices also affect the macro-
economy, as oil and gas exports account for big part of the total exports in the Russian
economy.9 We include the oil price in order to take account of external oil price
shocks as a source of business cycles.
Substituting equation [4] in equation [5], we get:
[6] ( ) ( ) ( )O
t t t t t t t ty r e Rπ λ ωθ ωβ φ ωδ ω ηπ ωε ν= + ∆ − + − ∆ − ∆ + + +
This equation suggests that as the real interest rate increases, inflation goes down;
when the currency depreciates, inflation rises; when international reserves go up,
inflation can be lowered on the back of allowing a currency appreciation, following
the surge in capital inflows. If the coefficient associated with the change in reserves
comes out negative, then the main factor that could be driving inflation in Russia is
exchange rate depreciation.
9 The gas price in Russia’s European export market is set in line with crude oil price movements.
12
The Bank of Russia determines the interest rate, lowering it to discourage
currency appreciation, leading to higher inflation via monetary expansion, hence
establishing a trade-off between the exchange rate and inflation. The exchange rate
can be included as a determining factor in the central bank’s reaction function, as
suggested by Taylor (2001). The main issue in the literature on flexible inflation
targeting is to what extent the real disturbances preventing the stabilization of both
inflation and the output gap justify the “departure from complete and (immediate)
stabilization of inflation” (Giannoni and Woodford, 2005: 96). Given the rate of
inflation as in [6], the monetary authority can adopt a monetary policy rule. In an
open economy context, a modified version of the closed-economy Taylor rule can be
written for the unobservable equilibrium real interest rate as follows10:
[7] *( )t t t tr by c d eπ π= + − + ∆
Where b, c and d are parameters, measuring the magnitude of the response of the
monetary policy to output, inflation, and changes in exchange rate, respectively. In a
high-inflation open economy, it is important to monitor the real interest rate as
negative real rates are harmful for economic activity.
This reaction function suggests first that as inflation increases the real interest
rate increases, and secondly as the real interest rate increases, money growth declines
from equation [2], and hence inflation gets stabilised from equation [6]. Also as the
exchange rate is defined as the domestic currency price of a foreign currency,
exchange rate depreciation can produce an increase in the real interest rate. As
discussed earlier, the central bank has focused more on exchange rate targeting rather
than inflation targeting as an instrument of monetary control, so the inclusion of both
10 See, for example, Molodtsova, Nikolsko-Rzhevskyy and Papell (2008).
13
exchange rate and inflation in a monetary policy rule is justified in order to
disentangle their effect on the real interest rate.
III. Empirical Results
A. Data
As explained in section II, we start our empirical analysis in 1995, as this period
corresponds more truly to a stronger resolve both institutionally and de facto by the
monetary authorities to reduce inflation. Table 1 gives a summary statistics of Russian
inflation since 1995 where four periods are identified. All the results (statistics and
graphs) are computed in EViews and RATS econometric softwares.
This paper extends Granville and Mallick (2006) where an optimal monetary
rule was derived by minimising a loss function, but using monthly data for the period
February 1992 to February 2005, without considering real output and oil price effects
on the macro economy and taking the refinance rate as the proxy for the interest rate.
In this paper, we use the overnight inter-bank lending rate rather than the refinancing
rate. Vdovichenko and Voronina (2006) explain that the refinancing rate may not be
the best proxy for the interest rate and instead they use the interbank market interest
rate on one day credit and the Bank of Russia interest rate on overnight deposit. This
is a better proxy for the interest rate, because banks in Russia over big part of the
period under analysis could not borrow at the refinancing rate from the Bank of
Russia. As a result, the refinancing rate may not be viewed as an equilibrium market
rate. Also we look at the real interest rate behaviour, as we have included a proxy for
economic activity.
14
As the time series considered in our theoretical review of interrelationships are
the real interest rate, the inflation rate, the rate of change of the exchange rate, output
growth, oil price inflation and reserves growth, we include these same variables in our
empirical exercise. In this context, it is worth mentioning Starr (2005), who uses
quarterly data to show how monetary policy affects real economic activity in Belarus,
Kazakhstan, Russia and Ukraine and finds mixed evidence but with greater effects for
Russia, where a significant effect of interest rates on output is found. In our paper,
data at a monthly frequency are used between January 1995 and December 2007.
Monthly observations on the seasonally unadjusted consumer price index, CPI, were
compiled from Datastream and were de-seasonalised using the US Census Bureau’s
X12 seasonal adjustment method in Eviews. Inflation rates (INFt) at an annual rate are
calculated using this adjusted data.11 The ruble/US$ exchange rate and the overnight
money market interest rate (inter-bank lending rate) are also taken from Datastream.
This interest rate is likely to respond to the changes in macroeconomic variables. We
calculate the real interest rate (RIRt) using the Fisher identity. Money supply here
refers to the broad monetary aggregate (M2), and money growth refers to the annual
rate of growth of M2 12 Reserves data (LIRA) refer to gross international reserves
(official reserve assets in US$), compiled from the Bank of Russia’s statistical
11 Our inflation rate is calculated as follows: inflation=((cpi_sa/cpi_sa (-12))- 1)*100; To calculate the
consumer price index, Goskomstat, the Russian State Statistics agency collect prices of 400
representative goods and services from 350 cities including every capital city of the 89 regions each
month. Monthly price changes are then aggregated for each of the 89 regions, where the weights are
based on the structure of household expenditures for the region in previous year. See Gibson et al.
(2004), page 3.
12 For more details on the money supply definition and its changes over time, see Granville and Mallick
(2006).
15
resources. The real activity variable is proxied by the industrial production index (IP).
We have used the growth in IP to reflect the real output growth in the economy and it
is available at a monthly frequency.
B. The Results
A necessary but not sufficient condition for any possible cointegration is that each of
the variables should be integrated of the same order (more than zero) or that both
series should contain a deterministic trend (Granger, 1986). The test results are
reported in Table 2. The standard Augmented Dickey-Fuller unit root tests for
inflation, IP growth, and reserves growth reject the unit root null hypothesis at 5%
level of significance, while DF-GLS and KPSS tests for inflation and real interest rate
– which outperform the ADF tests – indicate the presence of non-stationarity. DF-
GLS test suggests that the real interest rate is I(1). The inflation time series being
integrated indicates the persistence of inflation. Hence it is important to know the
source of such persistence in inflation.
More importantly, the sample period appears to have numerous structural breaks
in the policy of the Bank of Russia: the period of fast disinflation in the early part of
the sample altered by a period of relatively low (but still two digit) and stable inflation
and gradual real appreciation of the ruble before the crisis of August 1998 followed
by a period of fast growth with a low real exchange rate and since 2000 a period of
increased emphasis on inflation reduction and a declared switch from the use of the
exchange rate as a nominal anchor to monetary targets. We therefore test for possible
regime switches/structural breaks in the sample using the Zivot-Andrews test (see
Figure 2), which clearly reflects a major break in 1998.
16
Further, the estimation of unrestricted VAR models with integrated variables is
problematic, as applying the conventional statistical theory is no longer appropriate.
Given that the time series involved are either I(1) or I(0), we can apply Johansen’s
cointegration test to find whether there is any cointegrated relation between the
variables. Thus we use the reduced form theoretical relations in section 2 to identify
the cointegrated relations and derive a vector error correction (VEC) model, which can
be used to examine the impact of exchange rate shocks.
Figure 1 plots the real interest rate (RIR), inflation (INF), the rate of change of
nominal exchange rate (DEXR), the growth of industrial production (GRIP), reserves
growth (DLIRA) and oil price inflation (OINF). A visual inspection of Figure 1
indicates that the variables appear to have structurally changed. Such apparent
changes in the time series process can result from events such as a financial crisis or
significant government policy shifts. In the case of Russia, the respective instances of
these events were the August 1998 default and the change from a money anchor to an
exchange rate anchor for the stabilisation programme. Due to the currency crisis in the
late 1990s creating breaks in the data, we have added three dummy variables with the
aim to capture changes in the exchange rate volatility for the outliers in 1998 (May
and September) and in 1999 September, as exogenous variables in the VAR to
overcome the normality problem in the data generating process. Similar dummies
have been used in Esanov, Merkl and Vinhas de Souza (2005).
Thus we estimate a VAR with 4 lags as found to be optimal by the Schwarz
information criterion, which in general outperforms other lag selection test criteria
suggesting five lags (see Table 3). Within such a VAR, we test whether there is any
cointegrating relationship between the five variables (with oil price inflation
considered exogenously along with three dummy variables) for the full sample using
17
the Johansen (1988) maximum likelihood method. We find the existence of two
cointegrating relations using a linear model (with intercept and trend). The presence
of two cointegrating vectors has also been checked with alternate assumptions. The
trace statistics (LR test) indicate two cointegrating equations at 1% significance level.
The fact that the null r=1 is rejected at a 1% significance level means that there exists
two meaningful long-run relations between these variables (see Table 4 and Figure 3).
The Trace test indicates the existence of two cointegrating relations and we have
identified the first relation to be an interest rate rule and the second vector to be an
inflation equation by imposing five restrictions. Here we discuss the restrictions that
are imposed to identify the cointegration vectors. Denoting the two long-run
equilibrium relationships by β1 = [β11, β12, β13, β14, β15]′ and β2 = [β21, β22, β23, β24, β25]′
respectively, we impose the following restrictions on the vector of variables [RIR, INF,
DEXR, GRIP, DLIRA] for identification: β1 = [1, β12, 0, β14, 0]′ and β2 = [0, 1, β23, β24,
β25]′. The restrictions on β1 indicate an equation for RIR. Theoretically the second vector
must be interpreted as an inflation equation in order to have a complete model in which
exchange rate generates inflation, which in turn influences RIR. The over-identifying
restriction is the coefficient associated with international reserves: β15=0, which can be
included in the second vector, but may not make good sense in the first vector. So the
hypothesis that international reserves can enter the second cointegrating relation, but
not the first one, cannot be rejected. The likelihood ratio (LR) test statistic for testing
the binding β restrictions (rank =2) is distributed as χ2(1) = 2.013 [0.156], which is
accepted. The LR test for these restrictions identifies both the cointegrating vectors.
The normalized cointegrating equations obtained from Johansen test in Table 4 can be
written as:
18
[8] (16.71) (7.645) (6.206)
(7.127) (8.608) (3.842) (4.952)
0.366 1.32 0.513 0.003
0.007 0.593 0.035 0.254 0.004
t t t t
t t t t t
RIR INF LEXR T
INF LEXR GRIP LIRA T
= − + − ∆ −
= + ∆ − − ∆ +
where T refers to a time trend. Figures in the parentheses under the estimated
normalized coefficients are t-values. All the variables included in the normalised
cointegrating relations are statistically significant, exchange rate changes have
significantly influenced inflation, thus influencing real interest rate in the first long-
run relation. The signs of the coefficients turned out to be as per the apriori
expectation in equations [6 and 7]. The above two long-run relationships can be
summarized as follows:
First, an increase (decrease) in inflation by 1 percent is associated with an
increase (decrease) in the real interest rate by about up to 1.32 percent in the sample
period. It should be noted that the relatively high coefficient of inflation partly reflects
the fact that inflation and nominal interest rates have been highly volatile during the
first part of the sample period.
Second, an increase (depreciation) in exchange rate by 1 percent is associated
with a reduction in the real interest rate of about 0.51 percent. Also as the exchange
rate depreciates, there is 0.59 percent increase in inflation from the second long-run
relation. Considering the two relations together, the real interest rate increases by 0.27
percent in response to a 1 per cent currency depreciation.
Third, from the second cointegrating relation, there is a trade-off between output
growth and inflation, which implies that an increase (decrease) in output growth of 1
percent is associated with a decline (increase) of the real interest rate of 0.05 percent
from the first relation. The negative coefficient associated with reserves growth
suggests that inflation should decline. But the persistent inflation indicates that
inflation has been driven by changes in the exchange rate towards depreciation.
19
Finally, the inclusion of a time trend in both relations suggests that the real
interest rate has significantly declined over the sample period, inflation appears to
have significantly increased over time, most likely due to the authorities’ resistance to
exchange rate appreciation, which would otherwise have occurred following the surge
in foreign exchange inflows.
Given these long-run relations, we need to formulate a short-run model to
examine the dominant effect of a shock to these variables. Further, as there are four
lagged variables for each series, the summary responses can be more insightful, and
we address this by employing generalised impulse response (GIR) analysis with the
VEC model. The GIR functions are not sensitive with respect to the ordering of
variables as in the Choleski decomposition (Pesaran and Shin, 1998). The impulse
responses following a shock to each of the endogenous variables are presented in
Figure 4. The results suggest that the exchange rate is the key driving factor for
inflationary pressure; and changes in the exchange rate contribute more to inflation,
which in turn adds to the variability of the real interest rate in the long-run.
Besides this traditional identification strategy, to further validate the results, we
adopt Uhlig (2005)’s new sign restriction method, which is robust to non-stationarity
of series including breaks. Also the zero long-run restrictions may appear very
stringent, as they are based on statistical properties of the data rather than serious
theoretical consideration. Within this framework, we identify three types of
underlying disturbances, namely an oil price, exchange rate and monetary policy (real
interest rate) shock. Given that oil prices can capture supply shocks, having obtained
the parameter estimates of the reduced form VAR, we impose three sign restrictions
as follows:
20
1. The oil price inflation does not decrease (>0) in response to its own
positive shock.
2. The changes in exchange rate will not decline (>0) in response to its
own positive shock, i.e. exchange rate depreciation.
3. The real interest rate does not decrease (>0) in response to a positive
exchange rate shock, as monetary policy will tighten to back up the
exchange rate. As in Figure 1, the real interest rate always remains
positive. This has also been observed in the cointegrating relations, as
discussed earlier.
These restrictions seem reasonable in the light of the observed pattern in the Russian
data. As the domestic currency depreciates, inflation could increase, while exported
goods become cheaper relative to imported goods, thus increasing demand for
domestic goods and thus higher industrial production. But we do not pre-judge this
outcome, as we would like this to be revealed from the impulse response functions.
Also exchange rate changes can either exhibit depreciation or appreciation. We
examine both types of exchange rate changes to establish any asymmetric behaviour
in responses. These restrictions help derive respective impulse vectors, which are
defined as innovations to the VAR system in response to a unit shock in each
disturbance. We keep those impulse vectors whose impulse response functions satisfy
the sign restrictions and discard the others.
We undertake the exercise with both exchange rate depreciation and
appreciation, as the Russian central bank has been struggling to find a balance
between inflation and nominal appreciation of the rouble. As in Figures 5 and 6, we
find that inflation responds positively to rouble depreciation, but as foreign reserves
growth has been significantly high, it puts pressure on the central bank to allow the
21
currency to appreciate. Thus we examine what would happen if the currency
appreciation shock occurs (see figure 6). As this has not been practised by the central
bank, the response does not appear to mirror the depreciation case. Rather the
response remains mixed across the board. The exchange rate shock does not obey the
restriction in the long-run, and it starts to depreciate after 7 months. This clearly
suggests that Russian monetary authorities have been following exchange rate
targeting to keep it depreciated that in turn keeps inflation persistent. We also check
for the robustness of our result by replacing international reserves with broad money
(see figures 7 and 8). The responses still remain similar, as in the recent years
international reserves make up the big part of the monetary expansion. Besides, there
seems to be a positive real effect of exchange rate shocks in the long-run as found in
the impulse responses for GRIP (see figure 5). This comes out clearly with
depreciation, but not with appreciation. Thus the central bank remains reluctant to let
the currency appreciate thus depreciation fuels inflation.
Overall, the exchange rate channel has played an important role in the conduct
of monetary policy in Russia. Since 2000, the Russian central bank made an attempt
both to delay (or even reverse) the real appreciation of the ruble by actively buying
foreign reserves in the open market, and to reduce inflation from double digit levels.
Clearly, these two conflicting goals, without an effective ability to sterilize the
inflows, can directly lead to inflationary pressures in the economy. In this paper, we
measured empirically the implicit policy that the Russian monetary authority has
pursued – real appreciation deceleration versus reduction in inflation – finding that
exchange rate shocks are crucial determinants of inflationary pressure.
22
IV. Conclusion
In the long-run, we find that while inflation and exchange rate changes have
significantly influenced the real interest rate, it is exchange rate changes rather than
reserves and output growth which have had the stronger long-run impact on inflation.
This is in contrast to studies including the period before 1995, where monetary
aggregates were the key factor determining inflation as for example in Esanov, Merkl
and Vinhas de Souza (2005). In addition, the finding that real interest rates react
positively to inflation suggests that the ‘Fisher effect’ does not seem to hold in the
long-run in a high-inflation transition economy. For the Fisher effect to hold, the real
interest rate should not change in response to changes in inflation, and, consequently
the inflation rate can have a one-to-one positive effect on the nominal interest rate.
This result further strengthens our finding that the exchange rate changes have been
the driving forces behind interest rate determination, confirming that the central
bank’s policy objective was to prioritise resistance to real exchange rate appreciation
over price stability. Within a high oil price environment, the monetary authorities
seem to have prioritised stabilizing the exchange rate in search of short run output
gains over bringing down .inflation. This explains in part why inflation has become
entrenched at a relatively high level. This policy of balancing the inflation target with
a contradictory exchange rate target seems futile, since the resulting higher inflation
in any case drives up the real trade-weighted value of the currency, producing exactly
the loss of competitiveness which the authorities want to avoid.
23
References
Ball, L. and N. Sheridan (2005). Does Inflation Targeting Matter? The Inflation-
Targeting Debate. B. Bernanke and M. Woodford. Chicago, University of
Chicago Press: 249-82.
Bauer, C., and Herz, B., (2007) Credibility of CIS Exchange Rate Policies – a
Technical Trader's View, Emerging Markets Review, 8(1): 50-66.
Basdevant, O. and S. G. Hall (2002). "The 1998 Russian Crisis: Could the Exchange
Rate Volatility have Predicted it?" Journal of Policy Modeling 24(2): 151-168.
Central Bank of the Russian Federation (2004). “Guidelines for the State Monetary
Policy in 2005” Approved by the Bank of Russia Board of Directors on
November 15, http://www.cbr.ru/eng/today/publications_reports/on_2005e.pdf
Choudhry, T. (1998). "Another Visit to the Cagan Model of Money Demand: The
latest Russian Experience." Journal of International Money and Finance 17(2):
355-376.
Cotarelli, C. and P. Doyle (1999). "Disinflation in Transition 1993-1997." IMF
Occasional Paper, No. 179.
Domaç, I., and E.M. Yücel (2005). What Triggers Inflation in Emerging Market
Economies? Review of World Economics, 141 (1): 141 – 164.
Erceg, C. and A.Levin (2003). “Imperfect credibility and inflation persistence”,
Journal of Monetary Economics, 50(4): 915-44.
Esanov, A., C. Merkl and L Vinhas de Souza (2005). “Monetary Policy Rules for
Russia”, Journal of Comparative Economics, 33 (3): 484-499.
Ferguson, N. and B. Granville (2000). "Weimar on the Volga: Causes and
Consequences of Inflation in 1990s Russia Compared with 1920s Germany."
Journal of Economic History 60(4): 1061-1087.
24
Frankel, J. and M. P. Taylor (1993). "Money Demand and Inflation in Yugoslavia,
1980-89." Journal of Macroeconomics 15: 455-481.
Giannoni, M. P. and M. Woodford (2005). Optimal Inflation-Targeting Rules. The
Inflation-targeting debate. B. Bernanke and M. Woodford. Chicago and London,
National Bureau of Economic Research, Studies in Business Cycles, The
University of Chicago Press. 32: 93-172.
Gibson, J., S. Stillman and T. Le (2004). "CPI bias and Real Living Standards in
Russia during the Transition." William Davidson Institute Working Paper
684(May).
Granger, C. W. J. (1986). "Developments in the study of cointegrated economic
variables." Oxford Bulletin of Economics and Statistics 48: 213-228.
Granville, B. (1995). The Success of Russian Economic Reforms published by RIIA
distributed by Brookings, 1995.
Granville, B. (2001). The Problem of Monetary Stabilisation. Russia's Post-
Communist Economy. B. Granville and P. Oppenheimer. Oxford, Oxford
University Press, 93-130.
Granville, B. and S. Mallick (2006). "Does Inflation or Currency Depreciation Drive
Monetary Policy in Russia?" Research in International Business and Finance,
20 (2): 163-179.
Keller, P. M. and T. Richardson (2003). "Nominal Anchors in the CIS." IMF Working
Paper WP/03/179.
Kutan, A.M. and J.C. Brada (2000) “The Evolution of Monetary Policy in Transition
Economies”, Review, The Federal Reserve Bank of St. Louis, 82 (2),
(March/April 2000), pp. 31-40.
25
Molodtsova, T., A. Nikolsko-Rzhevskyy and D.H. Papell (2008). Taylor rules with
real-time data: A tale of two countries and one exchange rate. Journal of
Monetary Economics, 55(1): S63-S79.
Oomes, N., Ohnsorge, F. (2005). Money demand and inflation in dollarized
economies: The case of Russia, Journal of Comparative Economics, 33 (3):
462-483.
Papazoglou, C. and E. Pentecost (2004). "The Dynamic Adjustment of a Transition
Economy in the Early Stages of Transformation." Journal of Macroeconomics
26(3): 547-561.
Pesaran, H.H. and Y. Shin (1998). ‘Generalized Impulse Response Analysis in Linear
Multivariate Models’, Economics Letters, 58, 17-29.
Phylaktis, K. and Taylor, M.P (1993). ‘Money Demand, the Cagan Model and the
Inflation Tax: Some Latin American Evidence’, The Review of Economics and
Statistics, 75 (1), 32-37.
Starr, M. A. (2005). "Does money matter in the CIS? Effects of monetary policy on
output and prices." Journal of Comparative Economics 33: 441-461.
Taylor, J. B. (1993). “Discretion versus policy rules in practice.” Carnegie-Rochester
Conference Series on Public Policy 39: 195-214.
Taylor, J. B. (2001). "The Role of the exchange Rate in monetary policy rules." The
American Economic Review, Papers and Proceedings 91(2): 263-267.
Taylor, M. P. (1991). "The Hyperinflation model of money demand revisited."
Journal of Money, Credit, and Banking 23: 327-351.
Uhlig, H. (2005). What are the effects of monetary policy on output? Results from an
agnostic identification procedure, Journal of Monetary Economics, 52 (2): 381-
419.
26
Vdovichenko, A.G. and V.G. Voronina (2006). "Monetary policy rules and their
application in Russia." Research in International Business and Finance 20 (2):
145-162.
Wynne, M. A. and E. K. Kersting (2007). "Openness and Inflation." Staff Papers
Federal Reserve Bank of Dallas 2(April).
27
TABLE 1-Russian Inflation, 1995-2007: Summary Statistics
(annual percentage change in CPI)
Sample: Jan. 1995-July 1995 Aug. 1995-July 1998 Aug. 1998-Dec. 1999 Jan. 2000-Dec. 2007
Mean 219.7 49.5 81.2 13.3
Median 221.2 19.0 83.9 12.5
Maximum 225.7 222.0 125.7 25.9
Minimum 209.7 5.8 9.3 7.1
Std. Dev. 5.8 60.5 34.9 4.2
Observations 7 36 17 96
Source: calculated with data from Datastream
TABLE 2-Testing for stationarity
ADF(τ) ADF(µ) DFGLS KPSS
INFt -3.934** -3.757** -2.190 0.107 RIRt -5.208** -4.581** -1.127 0.248**
∆LEXRt -2.696 -2.505 -2.159 0.063
GRIPt -4.011** -3.742** -3.319* 0.092 OINFt -3.238 -3.122* -3.199* 0.047
∆LIRAt -2.548 -1.681 -2.929 0.075
5% -3.442 -2.881 -2.989 0.146
Notes: ADF(τ) and ADF(µ) are tests of the unit root null hypothesis, containing a constant and a trend, a constant and no deterministic components, respectively. Kwiatkowski-Phillips-Schmidt-Shin (KPSS) test assumes that the null-hypothesis is stationary. DFGLS and KPSS tests also include a constant and a
trend; * and ** denote 5% and 1% levels of significance respectively. INFt, RIRt, ∆LEXRt, GRIPt,
OINFt and ∆LIRAt are inflation, real interest rate, change in exchange rate, growth of industrial production, oil price inflation and change in international reserves respectively.
TABLE 3 -VAR Lag Order Selection Criteria
Lag LogL LR FPE AIC SC HQ
0 -216.4847 NA 2.83e-05 3.7151 4.2666 3.9392
1 679.2332 1653.633 4.31e-11 -9.6805 -8.5776 -9.2324
2 745.5728 117.3701 2.29e-11 -10.3165 -8.6621 -9.6443
3 789.7993 74.8447 1.71e-11 -10.6123 -8.4065 -9.7160
4 868.9578 127.8714 7.54e-12 -11.4455 -8.6883* -10.3251*
Notes: Sample: 1995M01 2007M12; Included observations: 130; Endogenous variables: RIR INF
GRIP, ∆LEXR ∆LIRA; Exogenous variables: C, DUM, DUM1, DUM2, OINF; * indicates lag order selected by the criterion; LR: sequential modified LR test statistic (each test at 5% level); FPE: Final prediction error; AIC: Akaike information criterion; SC: Schwarz information criterion; HQ: Hannan-Quinn information criterion.
TABLE 4: Unrestricted Cointegration Rank Test (Trace)
Hypothesized Trace 0.05
No. of CE(s) Eigenvalue Statistic Critical Value Prob.**
None * 0.713 252.514 88.804 0.000
At most 1 * 0.271 85.105 63.876 0.000
At most 2 0.169 42.749 42.915 0.052
At most 3 0.122 17.871 25.872 0.353
At most 4 0.004 0.490 12.518 1.000
Notes: Linear deterministic trend is included. Trace test indicates 2 cointegrating eqn(s) at the 1% level; * denotes rejection of the hypothesis at the 0.05 level; **MacKinnon-Haug-Michelis (1999) p-values
28
FIGURE 1 - Real Interest rate (RIR), Inflation (INF), rate of change of nominal
exchange rate (DEXRATE), Growth of industrial production (GRIP), Oil price
inflation (OINF) and international reserve growth (DLIRA), January 1995 to
December 2007
0.0
0.4
0.8
1.2
1.6
2.0
95 96 97 98 99 00 01 02 03 04 05 06 07
RIR
0.0
0.2
0.4
0.6
0.8
1.0
1.2
95 96 97 98 99 00 01 02 03 04 05 06 07
INF
-.2
.0
.2
.4
.6
.8
95 96 97 98 99 00 01 02 03 04 05 06 07
DEXRATE
-0.8
-0.4
0.0
0.4
0.8
1.2
95 96 97 98 99 00 01 02 03 04 05 06 07
DLIRA
-20
-10
0
10
20
30
95 96 97 98 99 00 01 02 03 04 05 06 07
GRIP
-0.8
-0.4
0.0
0.4
0.8
1.2
95 96 97 98 99 00 01 02 03 04 05 06 07
OINF
Source: Calculated using data from Datastream
29
Figure 2: Zivot-Andrews Structural break test for RIR
1997 1998 1999 2000 2001 2002 2003 2004 2005-5.6
-5.2
-4.8
-4.4
-4.0
-3.6
-3.2
-2.8
Figure 3: Estimated cointegrating relations
-1.2
-0.8
-0.4
0.0
0.4
0.8
95 96 97 98 99 00 01 02 03 04 05 06
Cointegrating relation 1
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
95 96 97 98 99 00 01 02 03 04 05 06
Cointegrating relation 2
Figure 4: Impulse responses from the vector error correction model
-.05
-.04
-.03
-.02
-.01
.00
.01
.02
.03
5 10 15 20 25 30 35
INF
GRIP
DEXR
DLIRA
Response of RIR to Generalized One
S.D. Innovations
-.08
-.06
-.04
-.02
.00
.02
.04
5 10 15 20 25 30 35
INF
GRIP
DEXR
DLIRA
Response of INF to Generalized One
S.D. Innovations
30
Figure 5: Impact of exchange rate depreciation, oil price inflation and RIR, on CPI Inflation
IRA and IP growth
Shocks in RIR, exchange rate depreciation and Oil inflation
Impulse Responses w ith Pure-Sign Approach
Impulse Responses for rir
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
Impulse Responses for inf
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
0.00
0.01
0.01
0.01
0.02
0.03
0.03
0.04
Impulse Responses for dlexr
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.01
0.00
0.01
0.02
0.02
0.03
0.04
0.05
0.06
0.06
Impulse Responses for grip
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1.60
-1.20
-0.80
-0.40
-0.00
0.40
0.80
1.20
Impulse Responses for dlira
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.07
-0.05
-0.02
0.00
0.03
0.05
Impulse Responses for doil
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.01
0.00
0.01
0.02
0.04
0.05
0.06
0.07
0.08
Figure 6: Impact of exchange rate appreciation, oil price inflation and RIR, on CPI Inflation
IRA and IP growth
Shocks in RIR, exchange rate appreciation and Oil inflation
Impulse Responses w ith Pure-Sign Approach
Impulse Responses for rir
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
Impulse Responses for inf
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.04
-0.03
-0.02
-0.01
-0.01
-0.00
0.01
0.01
0.02
Impulse Responses for dlexr
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.06
-0.05
-0.03
-0.02
0.00
0.02
0.03
0.05
Impulse Responses for grip
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1.50
-1.00
-0.50
0.00
0.50
1.00
1.50
2.00
Impulse Responses for dlira
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.07
-0.05
-0.02
0.00
0.03
Impulse Responses for doil
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
31
Figure 7: Impact of exchange rate depreciation, oil price inflation and RIR, on CPI Inflation,
M2 and IP growth
Shocks in RIR, exchange rate depreciation and Oil inflation
Impulse Responses w ith Pure-Sign Approach
Impulse Responses for rir
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0 .01
0 .00
0 .01
0 .02
0 .03
0 .04
0 .05
0 .06
Impulse Responses for inf
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0 .01
-0 .01
0 .00
0 .01
0 .01
0 .01
0 .02
0 .03
0 .03
0 .04
Impulse Responses for dlexr
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0 .01
0 .00
0 .01
0 .02
0 .02
0 .03
0 .04
0 .05
0 .06
0 .06
Impulse Responses for grip
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-1.50
-1.00
-0.50
0.00
0.50
1.00
Impulse Responses for dlm2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.02
-0.01
0.00
0.01
0.02
0.03
Impulse Responses for doil
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.01
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Figure 8 - Response to Exchange rate appreciation, RIR and oil price inflation
Shocks in RIR, exchange rate appreciation and Oil inflation
Impulse Responses w ith Pure-Sign Approach
Impulse Responses for rir
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0 .02
-0 .01
0 .00
0 .01
0 .02
0 .02
0 .03
0 .04
0 .05
Impulse Responses for inf
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0 .04
-0 .02
-0 .01
-0 .00
0 .01
0 .02
Impulse Responses for dlexr
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0 .05
-0 .04
-0 .02
0 .00
0 .02
0 .04
0 .05
Impulse Responses for grip
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-2.00
-1.50
-1.00
-0.50
0.00
0.50
1.00
Impulse Responses for dlm2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.04
-0.04
-0.03
-0.03
-0.02
-0.01
-0.01
-0.00
0.00
0.01
Impulse Responses for doil
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0.05
-0.03
0.00
0.03
0.05
0.08
0.10
32
Appendix 1: Variable Definitions
Variables Definitions Variables Definitions
Theoretical Empirical
supply of money M2
R international reserves
IRA international reserves
DA domestic assets
y level of real output IP industrial production index
GRIP growth of industrial production
r real interest rate RIR real interest rate
e exchange rate EXR nominal exchange rate
Md demand of money
parameters
expected inflation INF inflation rate
OINF oil price inflation
i nominal interest rate
l coefficient
f coefficient
b parameter
c parameter
d parameter
s
tM
δβθ ,,
εt demand disturbance
υt supply disturbance
πO oil price inflation
πe