T Mboweni: Star Safmarine Breakfast

7 September 2006

Honoured guests
Ladies and gentlemen

Introduction

Central banking has come a long way since the days, not too many years ago,
when monetary policy was shrouded in secrecy. Central bank communication was
akin to deciphering an obscure biblical tract, and monetary policy very often
surprised the markets. It is almost unbelievable that as recently as 1994 the
United States Federal Open Market Committee (FOMC) did not even announce the
monetary policy decision, let alone the justification thereof, and it was left
to the market to infer from the Fed's actions in the market whether or not
there had been a change in the monetary policy stance. The monetary economist,
Karl Brunner, was most scathing in his criticism and in 1981 argued that the
peculiar mystique surrounding central banking thrived on the pervasive
impression that central banking was an esoteric art confined to the initiated
elite. He further added that "this esoteric nature of the art is moreover
revealed by an inherent impossibility to articulate its insights in explicit
and intelligible words and sentences".

As we know, things are very different today. The pendulum has swung in the
opposite direction, prompting Alan Blinder, a former Vice-Chairman of the
United State Reserve System, to refer to it as the 'quiet revolution'. Much of
the debate about communication now focuses on the issue of whether central
banks can be too transparent. Notable economists in the field such as Blinder
and Lars Svensson argue for maximum transparency, while others, for example
Frederic Mishkin, argue that there are limits to transparency. I will argue
today that transparency and communication are a function of the decision-making
process and particular institutional features, which in effect means that no
single approach can be regarded as 'best practice'.

In looking at issues such as transparency and communication, the point must
also be made that communication is a two-way street. Making monetary policy
also involves giving attention to signals coming from the market. Indeed, the
argument is sometimes made that the central bank should simply follow the
market, raising the question of who takes the lead.

Communication from the market to the Bank

It is generally agreed that there is much market information of a
forward-looking nature that is extremely useful for monetary policy. Sometimes
central banks will follow market movements which can provide information about
the outlook that is independent of policy and that is crucial to monetary
policy decisions. Prices and interest rates incorporate all information
available to the market, and therefore provide important signals about the
future. True, there are times when the central bank may have information that
is not in the public domain, but equally, many private sector institutions are
privy to certain information, or may have superior information because of
specialised research in a particular area. Hence a two-way flow of information
does exist in practice.

The Bank therefore takes market information very seriously and we look at a
number of forward-looking indicators. These include the forward rate agreements
(FRAs), the yield curves and yield spreads, break-even inflation rates, the
Reuters consensus forecasts and the implied forward exchange rate curves, to
name a few. We are also in the privileged position to receive a wide range of
detailed in-house research from domestic and international financial
institutions. These give us some indication of market expectations concerning
various variables.

There is a rich literature on the information content of yield curves. The
generally accepted wisdom is that monetary policy determines the very short end
of the curve to a significant degree. However monetary policy does not
determine the long-term rates. These are determined by a number of variables,
including real output growth and the market's expectation of long-term
inflation and expectations of the future monetary policy stance. For this
reason there is not usually a close relationship between long-term bond rates
and the repo rate, although the yield curve may contain useful information
about the stance of monetary policy, and the market's expectations of future
inflation and real activity.

Interpreting the information contained in the yield curve or the yield
spread is not always straightforward. Standard theory tells us that the
long-term nominal rate reflects the real rate and inflation expectations. But
neither of these components is observable, and there may be other components as
well such as liquidity premia which complicate the story. Also, the extent to
which the long-term rates react to central bank actions depend on the
credibility of monetary policy. The information content of the yield curve or
yield spread is therefore not always clear-cut. I am sure you all remember not
too long ago Chairman Greenspan's long-term interest rate 'conundrum' in the
United States. Similarly, in South Africa, we were recently told that long-term
interest rates were being affected by liquidity considerations, as government's
fiscal prudence had resulted in a shortage of scrip in the market.

We would therefore not want to set policy on the basis of a single
indicator, and we would look to other indicators as well for corroboration.
Inflation expectations are critical to monetary policy decisions. Here we would
rely on a number of indicators, including the long-term bond yields, the yield
differential between conventional and inflation-linked bonds, the Reuters
consensus forecasts, and the inflation expectations survey that we have
commissioned the Bureau for Economic Research to conduct on our behalf. All of
these measures are imperfect, but they are indicative and, unless they
contradict each other, they could contain valuable information for policy
purposes.

Although there are strong arguments for taking note of market signals, this
does not solve all our problems. No matter how well-informed markets are, or
central bankers for that matter, they do not have perfect foresight. As we
know, the future is inherently unpredictable. Even if we make policy optimally
on the basis of all available information, new information becomes available
all the time. Behavioural relationships by their very nature are very difficult
to predict. For example, we may expect on the basis of our models that if real
incomes rise, consumer expenditure will increase in proportion to the estimated
parameters. However, certain other unpredictable factors such as natural
disasters, or geopolitical influences to name a few, are not easily
forecastable. As these events unfold and are absorbed by the markets, current
prices adjust, making previous forecasts redundant. This is the nature of asset
markets.

For this reason forward prices are changing all the time, which makes it
difficult to easily use all the market information with total confidence. It
also explains why forecasts, including our own, are periodically revised. Take
the example of oil prices. As you know, oil prices are an important factor of
our inflation rate. The international oil price is an exogenous variable in our
inflation forecasting model, but the inflation forecast is critically dependent
on our assumptions about the future path of oil prices. We are not oil market
experts, but we try to understand the supply and demand factors that impact on
the market.

We also look at the forecasts of a number of institutions, many of which
would have specialised oil price models and therefore presumably have a
superior view of the oil price outlook. However, these forecasts can vary
significantly. According to the Reuters consensus survey, at the June meeting
of the Monetary Policy Committee (MPC), the forecasts for Brent crude for 2006
ranged from US$56 to US$67 per barrel, and the range for 2007 was US$46 to
US$64. According to the latest Reuters consensus survey, the range for 2006 is
now between US$63 and US$72 per barrel, and for 2007 the range is between US$53
and US$75 per barrel. This indicates the enormous degree of uncertainty even
among institutions that put a lot of effort into forecasting the oil price. We
can take advice from the market, but whose advice do we take in this instance?
The average or 'consensus' is not necessarily superior to any of these
individual forecasts.

Some central banks simply look at the oil futures prices, rather than trying
to predict the impossible. But as we all know, the futures curve shifts
continuously as current prices change. All it tells us is what today's price
is, and what the implied price is over the next few months on the basis of a
risk-free interest rate and, storage costs. A sudden incident in Nigeria, Iraq
or Iran, or a new tropical storm system, for example, will suddenly change both
the spot and futures prices.

Similar reasoning can be applied to forward rates in the currency and
interest rate markets. These change almost continuously as new information is
absorbed by the market. The FRAs, for example, change every time there is a
change in the exchange rate, because of the expectation that this could lead to
a change in interest rates. As most studies show, even though forward rates,
whether currency, commodity or interest rates, are supposed to be unbiased
predictors of future spot rates, they are nevertheless not very good
predictors.

This is the uncertain environment in which we as the central bank have to
conduct monetary policy. Clearly, we cannot change monetary policy on a daily
basis. We have to take a forward-looking, longer-term perspective on the basis
of ever-changing data and events, news and surprises. While we can rely on the
market for some information, in the end we have to make the decisions on the
basis of our judgement. The collective view of the market helps us to make as
informed a judgement as possible.

Should the Bank follow the market in setting interest rates?

So far I have considered the issue of the flow of forward-looking data from
the market to the Bank. As I noted earlier, there is a strong case for the Bank
to use such information as part of its data for policy decisions. I do not
think that this is particularly controversial. A related, but more complicated
issue is the question of whether the Bank should follow the market when it
comes to setting interest rates. In other words, should monetary policy simply
set interest rates on the basis of market expectations?

Today it is generally accepted that monetary policy should be sufficiently
transparent so that the market can discount changes in monetary policy in
advance. If the market correctly anticipates monetary policy actions, there
will be little reaction to monetary policy announcements. One way of achieving
this favourable outcome is to simply follow market expectations.

However, if there is effective communication, market expectations of central
bank actions will not be independent of Bank actions and signals. Indeed, an
important dimension of monetary policy is to influence market expectations.
Under such circumstances, interest rate expectations are not formed
independently of the view that the market has of central bank actions. In other
words, if we signal a change in the monetary policy stance, this will be
reflected in current market rates ahead of the actual change in the policy
stance. Under such conditions, although it could appear that we are following
the market, we are in fact leading it.

Blinder, in his book "The Quiet Revolution", also warns against central
banks becoming too respectful of markets. He argues that slavishly following
the market could lead to poor policy for a number of reasons. Firstly, there is
the problem of herd behaviour, which may or may not be rational, in speculative
financial markets. This results in overreactions to stimuli whereas monetary
policy makers need to proceed with caution and prudence. Secondly, speculative
bubbles are a fact of life. As Blinder graphically puts it, 'central bankers
must steadfastly resist such whimsy and inoculate themselves against the
faddish behaviour that so often dominates markets. That may be why central
bankers are not much fun at parties'. Finally, he argues that market traders
tend to have much shorter time horizons than central bankers. This is even the
case where long-term bonds are traded, but are in fact treated as shorter-term
instruments. The result is that by following the market, monetary policy could
land up having a short time horizon and be prone to overreaction. It is
important for monetary policy to maintain a focus on the medium term. We have
to see through the short term noise and not be blinded by it.

Communication from the Bank to the market

As I mentioned at the outset, increased transparency is a feature of modern
central banking. Although transparency is an essential part of the inflation
targeting framework, it is not unique to this framework. It is also the case
that communication strategies of central banks differ widely, implying that
there is no single blueprint for central bank communication. In many cases, the
nature of the communication is determined by the institutional environment.

I think it is safe to say that the record of transparency at the South
African Reserve Bank has improved significantly with the introduction of
inflation targeting. The adoption of inflation targeting in itself was an
important step towards increased transparency as it includes the announcement
of a clearly defined overriding monetary policy objective that is, to maintain
Consumer Price Index, excluding interest rates on mortgage bonds (CPIX),
inflation within the target range of six to three percent. Other elements of
improved transparency include the detailed monetary policy statement released
after each MPC meeting, the publication of the Monetary Policy Review, the
Monetary Policy Forums as well as the numerous speaking engagements undertaken
by myself and my colleagues. Of course transparency is a matter of degree, and
there are some areas in which you may feel we fall short, but this is an
evolving process.

Transparency can also be taken too far and there are also some areas of what
I would regard as spurious transparency. Take for example our inflation
forecast. In the Monetary Policy Review, we publish the graph of our most
recent inflation forecast with the assistance of the fan chart tool. In our
monetary policy statements we have taken to giving some idea of our inflation
forecast at each meeting. We do not however publish the exact numbers
pertaining to each point, although we generally indicate the peak and the end
point. Some analysts have criticised the Bank for not publishing every point
along the trajectory. I would argue that this is a case of spurious
transparency. It will not make any difference to the policy decision if the
outcome in three years time is 4,8 or 4,9 percent. What we focus on is the
overall general trajectory and its relationship to the target range. In any
event, because of the inherently uncertain nature of the forecast, it is
depicted in the form a fan chart which illustrates the uncertainties.

There is a more substantive and difficult issue related to transparency. In
publishing inflation forecasts, the question arises as to how to communicate
the path of interest rates on which the forecasts are conditioned. Three
different methods are used in practice, each with its own problems. We have
adopted an approach common to a number of central banks, of assuming an
unchanged monetary policy stance over the forecast period. In doing so, we are
not committing ourselves in advance to a particular interest rate path. At a
simplistic level, this approach could tell us that if the forecast is above the
target, then the stance should be tightened and vice versa. It does not tell us
how the forecast will change if interest rates are changed, although different
scenarios on different interest rate assumptions can be run. It also does not
tell us what the optimal interest rate path would be to keep inflation within
the target range over the forecast period. Thus, interest rates may not remain
unchanged over the forecast period if inflation is to be kept within the target
range. Nevertheless this approach does illustrate clearly the rationale for
changing interest rates or leaving them unchanged.

An alternative approach, which has been adopted by a number of central
banks, is to look at the market forecasts of future interest rates. It has been
pointed out by various analysts that this approach contains a problem of
circularity. As noted earlier, market rates are set to some extent on the basis
of what the central bank is expected to do, so there is nothing to pin down the
system, leading to indeterminacy of inflation.

The third option, favoured by a number of academics, is for the Bank to give
its forecast of the future path of interest rates. The central banks of New
Zealand and Norway follow this route. The difficulty here is that the public
has to understand that these paths are not unconditional commitments, but that
they can change if the facts change. Unless this is fully understood, it can
have an adverse impact on monetary policy credibility. There are other
practical problems with this approach. In particular, how do you get a
committee of eight people to agree on a path of interest rates over the next
three years, when it is sometimes exceedingly difficult enough reaching
agreement on the current move! It should be borne in mind that monetary policy
in New Zealand is made by the Governor alone and not by committee, which
reduces some of the practical difficulties associated with this approach.

Even if a full path is not specified, there is also the question of whether
or not the monetary policy committee should signal future moves or policy bias.
Here again practice varies from country to country, as the ability to signal
effectively between meetings is often determined by the nature of the
decision-making process in a country. Where the responsibility rests with a
committee and decision-making is by consensus, as is the case in South Africa,
clear signals or commitments between meetings becomes more of a communication
challenge.

There have been times when we have been criticised for "surprising" the
markets. It may well be that in some instances the communication could have
been clearer, but it is also the case that the market does not always fully
appreciate the conditionality of the signals. Let me give two examples.

At the June 2004 MPC meeting, interest rates were left unchanged, but we
were concerned about the risks to the inflation outlook. At the press
conference following the meeting I warned that the 'party is over'. This was
widely quoted in the press and taken as a signal that we had reached the trough
of the interest rate cycle. At the next meeting, we lowered interest rates by
50 basis points, which led to accusations that we had misled the markets.

There were a number of significant changes between the two meetings. These
included the improvement in inflation expectations and the appreciation of the
rand which contributed to a lower inflation forecast. Two lessons come out of
this. Firstly, any signal that is given is not an unconditional commitment.
These commitments are conditional upon things remaining the same or changing in
a particular anticipated direction. Quite clearly, the conditions did not
remain the same in this instance. The second related lesson is that markets
should not lose sight of the fundamentals. Instead of focusing only on what the
Bank says at a particular point in time, we should all keep track of how the
fundamentals are evolving. Policy should then be assessed on whether the move
was justified on the basis of the new information.

A more recent illustration was provided in April this year when we raised
the repo rate by 50 basis points, again taking some in the market by surprise.
However, the tone of the statement issued at the end of the previous meeting
was intentionally 'hawkish', in order to signal a possible rate hike at the
next meeting. The markets recognised the hawkish nature of the statement and
most of the analysts picked up on this. Yet when the increase came at the
following meeting, it took some in the market by surprise. Perhaps we did not
communicate clearly enough, but perhaps the market was also not listening
intently enough.

Recent economic developments

Before ending, I would like to highlight some recent economic developments.
In the past few weeks there have been a number of data releases which are of
relevance to the Bank. Unfortunately most of these releases have not contained
much good news, and these developments underline the risks to the inflation
outlook that we have been highlighting in the past few MPC meetings.

CPIX inflation for July was 4,9 percent, slightly up from 4,8 percent the
previous month. The main drivers were food and petrol prices. At these levels,
we are still well within the inflation target range of six to three percent. Of
concern, however, is the broad-based rise in producer price inflation which
measured 8,1 percent in July, the highest year-on-year rate of increase since
January 2003. Of particular concern as well is the 18,3 percent year-on-year
increase in prices of agricultural goods.

Credit extension numbers show that consumers are still borrowing at a strong
pace and there are no clear indications that demand for credit is being
affected by the recent changes to interest rates. We are aware however that
there is likely to be a lag before these effects are seen. Twelve month growth
in bank loans and advances to the private sector accelerated from 23,3 percent
in June 2006 to 24,6 percent in July. Mortgage advances were a major
contributor to this.

Trade account data also indicate the continued underlying strength of
domestic demand. South Africa's trade deficit which had narrowed in June
widened to R53,2 billion in July. In July, exports increased month-on-month by
3,3 percent while imports increased by 6,6 percent.

It is not all bad news however. Growth in real gross domestic product
increased at an annualised rate of 4,9 percent in the second quarter of this
year, compared to a revised rate of 4,0 percent in the first quarter. Although
real value added by the agricultural sector declined significantly, real value
added by the mining sector grew by 3 percent. This followed three successive
quarters of negative growth in this sector. The manufacturing sector, which
experienced annualised growth of 6,1 percent show signs of sustained
resilience. The construction sector remains the strongest growing sector in the
economy.

Let me conclude by saying that communication between the Bank and the market
goes both ways. However, communication can only be effective if we listen to
each other. The Bank recognises the importance of effective communication and
we will continue to try and improve on this. We should also guard against
excessive communication, however, as too much information could also result in
excessive noise or confusion. Transparency does not however mean that we will
always be telling the markets what our monetary policy decisions will be in
advance of our meetings. The essence of transparency is for us to react in a
consistent manner in response to changes in the economy which impact on our
stated objective.

Thank you very much.

Issued by: South African Reserve Bank
7 September 2006