Monetary Conditions Index

1. Introduction
Monetary policy affects economic activity and
inflation through a series of channels, which are
collectively known as the transmission
mechanism. Changes in the monetary
authority’s policy rate are generally transmitted
into changes in the market and retail interest
rates, which can affect households’
consumption and saving decisions, firms’
investment and borrowing behaviour and finally
output and inflation. In a flexible exchange rate
economy, changes in the policy rate also affect
the value of the domestic currency vis-à-vis
other currencies, influencing the
competitiveness of domestic exports and
imports, and ultimately affecting net trade and
hence aggregate demand. In addition to this,
exchange rates can also have a direct effect
on consumer price inflation, via domestically
consumed imported goods.

The Monetary Conditions Index (MCI) was
developed in the early 1990s with the aim of
providing information on the stance of monetary
policy taking into account both the interest rate
and the exchange rate channels. It is a
weighted average of the short-term interest rate
and exchange rate. Initially it was used as an
operational target by the Bank of Canada and
Reserve Bank of New Zealand, but
subsequently its role diminished due to
problems in its construction and interpretation,
and it is now used less frequently and only as
one indicator amongst many in monetary policy

With this in mind, it is not the aim of
this paper to contribute to the current
conjunctural monetary analysis, but rather to
discuss the origins of the MCI and to highlight
some limitations and issues relating to its
implementation and use.

With the increasing financial complexity of the
modern economy, growing attention is paid to
how other financial variables including the price
of various asset classes affect the economy.
Moreover, policy makers have placed an
increased emphasis on financial stability
considerations given that changes in financial
variables affect wealth and balance sheet
considerations of various sectors in the
economy. This has led to an interest in the
development of Financial Conditions Indices
(FCI), which seek to provide a simple measure
of how financial market variables impact on the
economy above and beyond the standard
interest rate and exchange rate channels.

However, in many senses the FCI can be seen
as an extension of the earlier MCI. Moreover,
many of the methodological difficulties
associated with the construction of the MCI, as
well as many of the caveats and criticisms, are
also germane to FCIs.

The article proceeds as follows: Section 2
describes the MCI and its construction in more
detail, and identifies possible uses for the
index. Section 3 outlines the various important
issues related to the index regarding both its
methodology and its interpretation, while
section 4 looks at the MCI for some major
economies, and briefly discusses its
movements over the past decade. This article
also includes two boxes: the first provides
greater technical information on the choice of
weights used to estimate unobservable
elements of the MCI, while the second box is
based on a case study of the Canadian and
New Zealand MCI.

2. A description of the MCI

The MCI, which was first developed by the
Bank of Canada in the early 1990s, is
calculated based on a weighted average of
changes in short-term interest rates and
exchange rates relative to some reference
period. It aims to provide information on the
economy and inflation for monetary policy
analysis. A change in the index indicates how
‘tight’ or ‘loose’ monetary conditions in the
economy are, relative to a certain reference

The most obvious benefits of the MCI are that it
is straightforward, easy to understand, and, in
the past, was seen as a better indicator than
just focusing on interest rates, given the role of
the exchange rate in the transmission
mechanism. Even though it was used by
central banks, international organisations, as
well as financial corporations in different ways
over the years, it has various shortcomings.

It is
difficult to operationalise given that it combines
a monetary policy tool (interest rate) and a
macroeconomic outcome (the exchange rate)
and a lot of judgement is required for its

A Discussion of the Monetary
Condition Index 70 Quarterly Bulletin 01 / January 10

2.1 Definition

The basic formula for the MCI is as follows:
MCI = −[θ1(Rt − R*) + θ2 ((et − e*) × 100)]
(Equation 1)

• Rt represents the level of the short term
interest rate, and et is the log of the
effective exchange rate at a particular
point in time t

. If e increases it implies
that the domestic currency is
appreciating. Either real or nominal rates
for each of these variables can be used.
Short-term money market rates are used,
as they are closely aligned to the policy
rate, and the decisions by monetary
authorities transmit quickly into these

• The asterisk denotes the reference level
of each of the respective variables. In
theory, the reference or base levels for
the variables should reflect ‘‘neutral’’
economic conditions, but in practice this
is difficult to operationalise, hence a
simple average over a period of time is
generally used3

. Rather than focusing on
the absolute levels of the variables,
changes in the variables with respect to
this base level are used. If the (Rt − R*)
or respectively the (et − e*) component is
positive it means that the current interest
rate, or respectively the exchange rate,
is higher than that observed on average
during the reference period.

• The weights applied to interest and
exchange rates, θ1 and θ2 respectively,
typically add up to unity. The ratio θ1/θ2
reflects the relative impact of the interest
rate and exchange rate on the economy
as measured by either aggregate
demand or prices, although the former
method appears to be much more 2

The short-term rate usually employed is the 3-month interbank
rate, and the effective exchange rate is a weighted average of
bilateral exchange rates against major trading partners.

difference in the log of the exchange rate is mulitplied by 100 in
order to express it as a percentage. 3 Ideally, optimal or equilibrium levels of the interest rate and
exchange rate could be used, estimated from a Taylor type rule or
an equilibrium exchange rate model, but in practice these are
exceptionally difficult to accurately estimate.
prevalent in the literature4

. Therefore, if
there is a rise of θ1 percentage points in
the interest rate, it will have the same
effect on the policy goal as a θ2 percent
appreciation of the domestic currency,
so that a larger ratio will mean a weaker
overall affect of the exchange rate in the

There are a number of possible
methods to derive these weights, which
are outlined in Box 1.
• Finally, a negative sign is usually
attached to the overall computation of
the index so that, when there is a
decline (increase) in the index, as
defined by Equation 1 above, it indicates
that monetary conditions have tightened
2.2 Possible uses

In the implementation of policy, monetary
authorities focus on a number of variables, from
the ultimate target (frequently inflation) at one
end of the spectrum to the policy instruments
(such as the short-term interest rate) at the
other. Due to long lags and the indirect
connections between the target and the
instruments, monetary authorities resort to
operational targets6
, information variables and
indicators that link the two.

These intermediate
variables or targets are closely linked to the
ultimate target and are influenced by changes
in the policy instrument (Freedman, 1994). The
MCI falls within this group of intermediate
measures, and can be used as an indicator or
operational target in the conduct of monetary

When the Bank of Canada developed the MCI
in the early 1990s, it was used as an
operational target in the design of monetary
policy, and was then subsequently used in the
same way by the Reserve Bank of New

4 Some commentators criticised the practice of deriving the weights
from an aggregate demand function when the overall target was
inflation, as was the case in Canada and New Zealand.

one of the reasons for focusing on aggregate demand is,
‘‘because it is the output gap, along with expected inflation, that is
the principal driving force behind increases and decreases in the
inflationary pressures and it is changes in aggregate demand that
are a key determinant of changes in the output gap’’ (Freedman,

5 The rationale for the negative sign is that tighter monetary
conditions generally bear down on activity levels and looser policy
generally does the reverse.

6 The operational target of monetary policy is an economic variable,
which the central bank aims to control by use of its monetary
policy instruments. It is the variable the level of which the
monetary policy decision-making committee of the central bank
actually decides upon in each of its meetings (Bindseil, 2004).

A Discussion of the Monetary
Condition Index

Quarterly Bulletin 01 / January 10 71

Box 1: Calculating the MCI weights
The MCI, as outlined in Equation 1, contains certain unobservable elements that
need to be estimated, namely the reference levels for the interest and exchange
rates, and the weights attached to deviations between these variables and their
respective reference levels. Overall, the size of θ1/θ2 should capture the effect of
percentage point changes in the interest rate relative to a percentage change in
the exchange rate and its accuracy is conditional on the particular model used
for estimation. This box reviews the different methods used in the literature to
estimate the weights.

Batini and Turnball (2002) posit that there are
three main methods for estimating the MCIs’

Single Equation based MCIs

One of the most common ways of deriving the
weights based on the above rationale involves
estimating an aggregate demand function,
similar to the following:

∆yt = α ∆ Rt + β ∆ et + x + error

Where ∆ is the first difference operator which
captures the change in the variable over time1
y is Gross Domestic Product (GDP), R is the
interest rate and e is the exchange rate.

subscript t refers to the current or latest time
period. Overall, this function seeks to discover
the effect of changes in interest rates,
exchange rates and other economic variables,
represented by x in the equation above (i.e.
current and lagged values for GDP of main
trading partners), on GDP.

From this equation,
α and β, the partial derivatives of the interest
and exchange rates respectively, can act as
the weights θ1 and θ2, so that the ratio of the
coefficients in equation 1, θ1/θ2, equals the ratio

Multiple Equation based MCIs

There are also more elaborate multiple
equation-based methods. These methods
involve estimating and simulating structural
macro-econometric models in which the
weights are then obtained from a system of
equations rather than just one.

The weights can
also very often be estimated using vector
autoregressive models (VARs), with time series
of GDP, exchange rates and interest rates.

Subsequently, impulse response functions
(IRFs) are derived. The IRFs measure the
response of GDP to individual shocks in both
the interest rate and the exchange rate.

weights θ1 and θ2 are then based on a
cumulative average responsiveness of GDP to
shocks in the interest and exchange rate
respectively over a certain number of quarters.

A critical element in the use of this approach is
the correct identification of shocks to the
relevant variables.

Many banks and
international organisations use the weights
estimated from existing structural
macroeconomic models, for example the OECD
bases its weights on results from their Interlink

MCIs based on large macroeconometric models, especially those that
contain a monetary policy reaction function, are
more instructive as they take account of more
features of the economy.

Trade share based MCI

This final method is simpler to calculate. The
exchange rate weight is based on the long run
exports-to-GDP ratio and the interest rate
weight is simply one minus this ratio.

rationale is that this net trade component
captures the effect of the exchange rate on
GDP relative to interest rates.

However, it is
used less frequently given the simplicity in
relation to the estimation of the weights and,
consequently, the lack of detail about the
effects of the relevant variables on the

Overall, the multiple equation based model is
generally deemed to be optimal, as it takes
account of the cumulative lagged impact of the
different variables.

The dynamics of the
underlying model are very important; a model
that takes account of the different lags at which
an economy responds to changes in interest
rates and exchanges rates would perhaps
deliver a more accurate index. If a model is too
simple, or fails to take into account key
characteristics of behaviour, the measurement
of the weights can be flawed, meaning that the
MCI itself is built on erroneous foundations.

The operator is defined as ∆yt = yt − yt−1, which is the change in
GDP from the previous period.

A Discussion of the Monetary
Condition Index

72 Quarterly Bulletin 01 / January 10
Zealand between 1997 and 1999 (see Box 2).

The rationale for adopting this policy was that it
may be difficult to predict the response of the
foreign exchange market to a change in the
policy rate (Gerlach and Smets, 2000).

theory of uncovered interest rate parity7
suggests that interest rates and exchange rates
are related in a systematic way, although
empirically this relationship does not always

Hence, there is still no completely clear
understanding of the interaction between
interest rates and exchange rates.

Above is wrong

The method used in the case of the operational
target, which was particular to these two
countries, involved having an inflation target8
and deriving a solution for future interest rates
and exchange rates consistent with the target
after having taken into account domestic and
foreign economic conditions.

Using these
projections the bank was able to derive the so called ‘desired’ MCI level, which could
represent a range of values rather than point

The forward-looking focus of this
approach took into account the lags between
the monetary policy stance or changes in it and
the effect on the rate of inflation9

. If the actual
level of the MCI deviated from the desired
path, the Bank would use the tools at its
disposal (for example, the overnight rate) to
adjust the index accordingly.

It is important to note that using the MCI as an
operational target does not imply an automatic
reaction to all exchange rate changes, since
the target level of the MCI varies in response to
shocks that affect the exchange rate.

In the
case of an aggregate demand shock, the
desired level of the MCI will change, whereas if
there is a credibility shock, the target MCI level
should remain unchanged ceteris paribus.

Charles Freedman, the deputy Governor of the
Bank of Canada at the time put it,
‘‘A lot of judgement goes into it, and there
is a lot of cross-checking against important
information variables such as the rate of
growth of monetary aggregates10’’.

7 The theory of uncovered interest rate parity states that ‘‘the
exchange rate against a foreign currency deviates from its
expected value at some future time by the size of the interest rate
differential (over the appropriate horizon) with that country’’
(Stevens, 1998).

8 Canada had an inflation range of 1-3 per cent and New Zealand
had a target of 0-3 per cent. 9 For Canada, Freedman (1995) estimated that monetary actions
would influence the rate of inflation in about 6 to 8 quarters ahead.

10 Excerpt from remarks made by Deputy Governor Freedman to the
Conference on International Developments and Economic
Outlook for Canada, 15 June 1995.

The use of the MCI as an operational target
diminished over time, due to pitfalls that
emerged when the index was used in this
capacity. In particular there was great
uncertainty regarding the source of exchange
rate movements. A more detailed look at these
problems in Canada and New Zealand, are
discussed in Box 2.

With the MCI’s relevance as an operational
target declining, it has increasingly been used
as an indicator in monetary policy analysis.

this capacity, monetary policy tools are no
longer used to adjust the level of the index to a
desired path, but rather it merely helps to
inform policy makers of the current stance of
monetary conditions, and whether they are
tighter or looser relative to other periods.

3. An evaluation of the MCI

The MCI presents some problems both at the
level of construction and in terms of its
conceptual and empirical foundations, which
are outlined in this section.

3.1 Methodological Issues in constructing

In constructing a MCI, an initial technical issue
is to determine the appropriate weights.

the weights of the components are not directly
observable, but are based on econometric
estimates, they are highly sensitive to the
model used (see Box 1) — i.e. the MCI ratio
can suffer from model uncertainty.

The main
pitfalls involved in deriving the weights
therefore, vary between the models used and
are consequently model dependent.

principal problems include capturing the
correct dynamics of the relationship, as interest
rates and exchange rates can affect the
economy at different speeds, and parameter
constancy, which requires that the coefficients
from the models used to calculate the weights,
must not change depending on the time period

11. If these problems are not adequately
dealt with, the weights that are derived risk
being erroneous and may provide an
inaccurate picture of monetary conditions (Eika
et al. (1996) and Batini and Turnball (2002)).

11 For a further and more detailed discussion of the econometric
problems involved in calculating the MCI weights please refer to
Eika et al. (1996) and Batini and Turnbull (2002).

A Discussion of the Monetary
Condition Index

Quarterly Bulletin 01 / January 10 73

Box 2: MCIs as an operational target — problems in practice

A number of difficulties and challenges emerged during the period in which the
MCI was used as an operational target by both the Bank of Canada (BoC) and
the Reserve Bank of New Zealand in the 1990s.

Some examples of these
problems are highlighted in this box.

While the MCI appeared to be attractive as an
operational target for the BoC, it became
evident that there were a number of

Firstly, there was a tendency on
the part of some observers to treat the MCI as
a precise short-term target for policy, while the
Bank indicated that it should not be treated as
a narrow, precise measure.

Secondly, the
markets started to treat all exchange rate
movements as portfolio readjustments on the
part of investors (portfolio shocks) and,
therefore, came to expect an offsetting interest
rate adjustment every time there was a
movement in the exchange rate, whether or not
such an adjustment was appropriate.

addition, the central bank itself had to make a
judgement on the source and likely persistence
of the shock to the exchange rate, in order to
decide on the appropriate response.

this caused problems in 1998, when the rapid
depreciation of the Canadian Dollar produced
accusations of a myopic central bank (Robson,
1998), whereas the BoC argued that the
depreciation signalled looser than desired
monetary conditions that warranted sharp
increases in the policy interest rate.

Given the difficulties mentioned, less emphasis
was placed on the role of the MCI as a
Related to this, even if the model for deriving
the weights is correctly specified and manages
to accurately capture the effects of the interest
rate and the exchange rate on the economy,
over a certain period of time, there is always
the possibility that the monetary transmission
mechanism itself (the effect of the interest rate
and the exchange rate on output) can change
over time for a variety of reasons.

Therefore it is
vital to monitor this system and ideally to
ensure that any changes to how monetary
impulses are transmitted to inflation are
recognised in the MCI. In practice this may be
difficult to achieve.

measure of monetary conditions in the late
1990s and the early part of the current decade.

Subsequently, the MCI was discontinued from
being published by the BoC (2006), and has
not been used as an input into monetary policy

Problems also emerged in New Zealand over
the period when the Reserve Bank of New
Zealand employed the MCI as an operational
target (mid-1997-March 1999).

In particular,
interest rates were increased as an automatic
response to a depreciation of the New Zealand
Dollar (NZ$), with little evidence that those
interest rate increases were warranted.

As a
result, interest rates were increased at a time
when a serious drought caused severe water
shortages in New Zealand, the Asian crisis
evolved (1997/1998) and as output growth in
the country turned negative.

Given the
circumstances, this may not have been the
most appropriate action.

Following the
difficulties encountered, the Reserve Bank of
New Zealand subsequently acknowledged that
they ‘‘were slow to recognise the joint impact of
the Asian crisis and the beginning of an
extended drought through 1997 and early
1998’’. They subsequently discontinued using
the MCI in this capacity.

A further technical issue is whether MCIs
should be calculated in terms of real or
nominal variables.

Theoretically, it would seem
preferable to express the MCI on the basis of
real variables as the real MCI takes account of
inflation movements. It is also generally
believed that rational agents consider the real
rather than nominal rates in their consumption
and investment decisions.

However, there is
evidence that individuals can suffer from
money illusion whereby they consider the
nominal rather than the real variables in their
decision making (Akerloff and Shiller, 2009)
(Fehr and Tyran, 2001). Peeters (1999) and
Gerlach and Smets (2000) also put forward the
A Discussion of the Monetary
Condition Index

74 Quarterly Bulletin 01 / January 10
argument that economic behaviour often reacts
on the basis of nominal interest rates in the
short run. Furthermore, the nominal MCI seems
to be a reasonable approximation for the real
MCI in the short run, in the context of a low
inflation environment. See Costa (2000b) and
the ECB (2002).

There are also other factors justifying the use of
nominal variables.

For example, a nominal MCI
may be easier to construct and is also timelier
as inflation data needed for the real measure
are only available on a monthly basis, as
opposed to the daily availability of nominal
interest and exchange rate data.

However, it
should be noted that in a period of high
inflation, the nominal index is likely to show
more pronounced tightening than the real

The selection of the MCI components is also
an issue that has received more attention in
recent years.

Since the MCI components
should be in line with the nature of the
monetary transmission mechanism and with the
appropriate structure of the relevant economy,
it has been argued that other factors, such as
long-term interest rates12 and asset prices (i.e.
house prices and stock prices), should also be
included in the MCI.

For the euro area, long term interest rates play an important role in the
monetary transmission mechanism, as
investment and consumption behaviour is often
dependent upon long-term rates.

Taking into account the increasing debate over
the role played by asset prices in the monetary
transmission mechanism, through wealth
effects and balance sheet effects, the Financial
Conditions Index (FCI) has been developed in
recent years.

Policy makers and international
organisations often use the FCI in their
assessment of the monetary policy stance.

However, the definition of FCIs differs across
methodologies. While some researchers
compute FCIs that measure the
tightness/accommodativeness of financial
factors relative to their historical average in
terms of an effective policy rate (e.g. Guichard

For countries in which long-term financing relationships play a
major role, it would be logically consistent to include a long-term
interest rate. Fixed long-term interest rates exert a larger
influence on consumption and investment decisions in several
countries in Continental Europe, relative to the Anglo-Saxon
countries (Costa, 2000).
and Turner, 2008),

others measure the
estimated contribution to growth from financial
shocks in a given quarter (Swiston, 2008).

The FCI extends the MCI approach by
including other financial variables, including
stock prices, asset prices and long-term
interest rates13, but similar to the MCI, the index
still suffers from certain criticisms, such as
model dependency, ignored dynamics and
parameter inconsistency.

While a full
discussion on such an index is beyond the
scope of this article, a number of interesting
findings are worth noting.

Based on research at
the BoC, which suggested that asset prices
may offer important information about future
inflationary pressures, Gauthier, Graham and
Liu (2004) estimate a number of FCIs for

They find that the FCI outperforms the
MCI in many areas, and also that house prices,
equity prices and bond risk premium, in
addition to short and long-term interest rates
and the exchange rate, are significant in
explaining output in Canada.

Goodhart and
Hofmann (2001), also find that house and share
prices are important variables in such an index
for G7 countries, and that the FCI contains
useful information about future inflationary

A more recent example of the FCI’s
use is illustrated in a recent paper by Beaton,
Lalonde and Luu (2009).

It looks at the
development and increasing importance of
financial conditions in the US during the current

They find that financial conditions have
had a large negative impact on US GDP
growth in the current recession14 and that the
monetary easing undertaken by the Federal
Reserve over the recent financial crisis has not
been sufficient to offset the tightening of
financial conditions.

A final point related to the selection of the
components is that neither money nor credit
plays a role in standard representations of a

For example, the same level of a MCI
could be consistent with various rates of
monetary growth, and in no way calls into
question the importance of the money supply in

More recently, lending standards have also been included in
FCIs to account for non-price credit conditions (Guichard and
Turner, 2008).

14 Their FCI suggests that financial factors subtracted between 4
and 7 percentage points from quarterly annualised growth in
2009 Q1.

A Discussion of the Monetary
Condition Index
Quarterly Bulletin 01 / January 10 75

the economy or ultimately the monetary nature
of inflation.

This is particularly relevant when
there are quantity constraints or credit
rationing. Given these shortcomings, the MCI
should be interpreted with caution regardless
of whether it is being used as an operational
target, or merely as one indicator among

3.2 Interpretation Issues

Irrespective of the difficulties in constructing a
MCI, interpreting changes in it in terms of their
significance for current monetary policy is not

Whether it is appropriate or not for a
central bank to make a policy change in
response to a change in the MCI (in the case
of the MCI being an operational target)
depends on the factors underlying changes in
the components.

A given movement of the MCI
may have different consequences in terms of
the final policy objective.

In particular, it is
important to determine the nature of shocks
causing movements in the exchange rate, and
not mechanically follow movements in
individual components, as highlighted in the
case of New Zealand in Box 2.

Siklos (2000) believes that the simplicity of
MCIs implies a loss of information when the
effects of the component variables are
aggregated, as it can obscure the movements
of the individual components.

King (1997) also
makes the point that ‘‘any attempt to construct
a simple monetary conditions index is akin to
adding together apples and oranges’’,
particularly given that the exchange rate is not
a policy instrument and therefore MCIs mix
variables that are not of the same nature.

Given the difficulty in determining whether any
particular reference period is ‘‘neutral’’ (Banque
de France Bulletin Digest, 1996), most
implementations focus on changes in the index
compared with previous periods, to ascertain if
monetary conditions have tightened or
loosened, rather than looking at the absolute
levels of the index.

It is important to note that
historical averages do not necessarily
represent neutral conditions and furthermore,
structural changes in the economy and
differences in cyclical conditions may also
affect what is understood as neutral conditions.

4. Constructing the MCIs

The following section focuses on developments
in MCIs for the euro area, the UK and the US
over the past decade.

It is important to
emphasise that the purpose of this section is
not to speculate as to which weights may be
best or even to assess the extent to which the
MCIs portray an accurate picture of the
monetary stance.

For each country, there are many plausible
alternative weights specified by various
institutions and academics.

The choice of the
weights can affect the overall level of the index
and also the rate of change of the indices.

this analysis, a weight of 6:1 is used for the
euro area, which is used by the European

and weights of 3:1 and 10:1 are
used for the UK and the US respectively.

latter weights have been applied by the IMF in
the past.

In terms of the data used in the
construction of the MCI for each country/area,
the short-term interest rate is proxied by the
three-month money market rate while a broad
trade weighted exchange rate proxies the
exchange rate variable.

These series are then
deflated by consumer price indices16.

the base/reference periods refer to the average
of both interest and exchange rates from 1993
to present.

4.1 MCIs in Practice

Referring to Chart 1, it is evident that over the
sample period (1999-2009) for the euro area,
there appears to have been a marked
tightening in monetary conditions post 2005,

which is consistent with increasing interest

Despite the significant economic
developments since the financial crisis, the
MCI, while volatile, has not shown any
substantial changes in its trend.

Meanwhile, the
MCI for the UK shows its only major shift in
trend from around 2007 on.

At this point both
interest rates and exchange rates contributed
to looser monetary conditions.

The movement
of the US MCI during the sample period has
been more variable, and has tended to track
changes in the interest rate, given this

15 This ratio was derived from simulations of the OECD Interlink
Model. 16 Ex-ante real short-term interest rates were also calculated (using
inflation 3-months forward), but the results were very similar.


Brian Twomey

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s