Monetary policy and its transmission channels in Mexico
Francisco Gil Díaz *
Introduction
This paper first deals generally with monetary policy and the various mechanisms by which policy
measures are transmitted to interest rates, the exchange
rate, the price level and output. It then uses this framework to present some features of contemporary Mexican monetary policy which has undergone a gradual process of adaptation since
1995, moving from the determination of central bank daily intervention interest rates, to a signalling mechanism based on very
small adjustments in the central bank’s provision of liquidity.
The paper begins with a description of the frequently misunderstood roundabout process of money creation. These misunderstandings are the origin of considerable confusion
about the role of a central
bank, the scope of its instruments and the way monetary policy
impacts on the economy.
The reference point for the discussion will be the widespread notion that central
banks can achieve
short-run increases
or decreases in the quantity of money.[1]
This notion is found in economics textbooks, in academic journals and in many non-specialised writings.[2]
Two recent examples, randomly
chosen, would be the following: “Tighter monetary policy: by end-December, the central
bank began to retire liquidity
from the economy, trimming
the monetary base by 4.9 billion
pesos between December 27 and January 3.”[3] Another statement,
of an opposite nature, that seems to focus correctly on the issues is: “Since there are sound reasons (i.e., Christmas) for output to leap in the fourth quarter, this suggests that output growth causes money supply growth, and not the other way around: as consumers
demand more money for their shopping, private and central banks increase the money supply.”[4]
The presentation of some general
principles of monetary policy in the next section will consider only some essential ingredients, so as to convey in
the simplest possible
manner the idea that, regardless of the institutional
arrangements (whether or not commercial banks are subject to required reserves,
or want to hold settlement balances), a central
bank has no possibility of immediately altering even the narrowest of the monetary aggregates:
the amount of currency in circulation.
As Goodhart (1987) puts it, customary textbook presentations misleadingly take the traditional
multiplier model as their starting-point. This approach contributes to propagating
the erroneous idea
that central banks
can alter the quantity
of money more or less as in the Patinkinesque money-from-
a-helicopter parable. Because
this thinking is so ingrained, the conceptual discussion will be elaborated in some detail.
The ultimate purpose of the presentation is to provide a framework
for understanding the signalling procedures utilised by the Banco de México to ease or tighten monetary conditions. The final part of the paper and the two annexes deal with
the Bank’s utilisation of its instruments in the past two years.
There is also a presentation of their impact on
key variables to illustrate how the policy signals are transmitted to interest rates
and the exchange rate, and from them to economic activity and prices.
Monetary policy principles
A
monetary authority
can make its credit dear or cheap and, in principle, it
can also issue currency or act upon the monetary base. But the fact that it
has the potential to issue currency does not mean it can do so at its discretion. It will be argued
that even in the medium
term, beyond a quarter or a year, for example, a central bank can increase or decrease the quantity of money
only in an indirect fashion and never with the mechanical,
immediate kind of result commonly portrayed as the outcome of open market
operations.
The central bank will be able to
modify the quantity of money only if it
is able to influence the demand for money. Furthermore, the actual change in the quantity
of money will take place only as the central
bank responds passively to accommodate changes
in the demand for money. An
increase or decrease
in the demand for money
may have been induced months (or years) before.
One of the indirect channels
for altering the quantity of money demanded is to influence
interest rates. The latter can change because of modifications in the policy stance of the central bank or because
international interest rate movements
influence local ones. Interest rate changes induced by the central
bank may act upon the exchange rate and upon aggregate demand. As they impinge
upon output and/or prices, the demand for money will be affected in turn.
Movements in the demand
for money may also be caused by non- monetary shocks such as an increase in the general price level brought about by a devaluation, perhaps provoked by a fall in the terms of trade or a
political disturbance. But a terms-of-trade shock is also a supply shock
and will cause changes in output that will influence
the demand for money as
well.
To
construct a schematic model useful for analysing these influences a few general concepts
are necessary.
The central concepts germane to the discussion are the public’s demand for financial assets and the fact that banks operate under a fractional
reserve system. Fractional reserve banking means that banks do not have the cash at hand to respond to variations in the public’s
demand for financial
assets, which occurs
daily as individuals’ preferences shift between cash and deposits
issued by commercial banks
* I would like to thank Guillermo Aboumrad, Agustín
Carstens, Pascual O’Dogherty, Moisés Schwartz and Abraham Vela for their helpul comments.
Responsibility for the opinions expressed in this paper is solely the author’s.
[1] The terms money and currency will be used interchangeably in this paper. Base or high- powered money can also be equated with currency because the payments
mechanism in Mexico allows banks to hold practically no settlement balances.
[2] With notable, perhaps not widely known exceptions, such as Goodhart
(1987) who
writes: “On this thesis,
the central bank undertakes open market operations, in order to vary its own liabilities, and, in the process,
the reserve base of the banking
system[...] In practice, however, the banking
system has virtually
never worked in that manner. Central
banks have, indeed, made use of their monopoly control over access to cash and their power to enforce that by
open market operations, but for the purpose of making effective a desired level of (short- term) interest rates, not to achieve a pre-determined quantity of monetary base or of some determined, variable is the change in (short-term) interest rates while both the monetary base and monetary aggregates are endogenous variables. This reality
is, unfortunately, sharply
in contrast with the theoretical
basis both of many economists’ models, and also of their teaching monetary aggregate[...] Indeed, central banks have historically been at some pains to assure the banking system
that the institutional structure is such that the system as a whole
can always obtain access to whatever
cash the system may require in order to meet its needs[...] In short, the behavioural process runs from an initial
change in interest rates, whether administered by a central bank or determined by market forces, to a subsequent readjustment in monetary aggre- gate quantities: the process does not run from a change in the monetary base, working via the monetary base multiplier, to a change in monetary aggregates, and thence only at the end of the road to a readjustment of interest rates. In reality, the more exogenous, or policy-determined, variable is the change in (short-term) interest rates while both the monetary base and monetary aggregates are endogenous variables. This reality
is, unfortunately, sharply
in contrast with the theoretical
basis both of many economists’ models, and also of their teaching.
[3] JP Morgan, Data Watch: México.10th January
1997, p. 9
[4] The Economist, 14th
– 20th, December 1996, p. 80.
Typically, there are seasonal movements between cash and deposits issued by commercial banks that occur within any given week, within any given month, and also within the year. Because of fractional reserve banking these shifts have to be financed, lubricated as it were, by the central bank. But this continuous and passive accommodation by the central bank of the public’s needs for cash does not mean that the central bank can go about injecting or withdrawing cash from the economy beyond what the public is demanding. Fractional reserve banking and the nature of the demand for currency do not allow it.
To
substantiate this argument, fundamental to the understanding of monetary policy, it will be shown first why the central bank cannot bring about systematic and instant
changes in the money supply. Afterwards, the diverse channels of money transmission and their implications for monetary policy
will be explored.
To
develop the argument it will be useful to start with a simplified accounting representation of the economy.
We
shall consider that the central
bank’s only assets
are credit to the commercial banks, A, and international reserves, IR. It has no capital and only one liability, represented by its own notes and coins (M0).
The only asset of the commercial banks is government debt, G. They also have no capital and on the liability side they issue deposits, D, held by the general public, in addition to the credit they receive from the central bank, A.
This skeleton model is sufficient
for the exercise that follows. Any more “realism” would not change
the outcome and would add unnecessary complications.
Consider an attempt by the central bank to reduce the supply of currency, M0. In principle
it can do so by trying to reduce the size of its credit to the commercial banks, A. In response
to the attempt by the central bank to reduce A, commercial banks would have to:
(a)
reduce G, or
(b)
reduce G and increase D, or
(c)
increase D.
G
cannot fall automatically as a consequence of monetary policy. Since a
central bank cannot determine fiscal policy or synchronise its own monetary policy with it, unfeasible alternatives (a) and (b) will be ignored.
Given the fixity of G, commercial banks have to maintain the level of their funding. Therefore, for the central
bank to be able to reduce its credit, A, to the commercial banks, it would have to induce an increase in the amount of deposits held by the public at commercial banks, D. An increase in D involves a very substantial rearrangement of the public’s portfolio, away from cash and into bank deposits. If this occurred, the amount of financial assets held by the public would remain the same, the reduction in credit to the commercial banks from the central bank would be
compensated by the rise in deposits from the public, and the central bank would be able to contract A and M0 by equal amounts. But since M0 is
needed for transaction purposes, the increase in the interest rate necessary to entice the public to instantaneously relinquish the required amount of M0 in order to hold more D would be outlandish. It is not realistic to assume that short-run, or even longer-term monetary policy could rely on such a
mechanism.[5]
A
strong initial conclusion is that the prevailing
idea, held by many economists and lay people, that the central bank can change the amount of
currency in circulation in the short run is simply wrong. To achieve this result we have been assuming, paradoxically, that the central bank can control the supply of currency, at least in principle. To be able to increase the
supply of currency someone has to be willing to demand the expanded amount.
A
change in the
quantity of money
brought about by the
central bank’s deliberate actions also requires a flexible
exchange rate and an adjustment period. Under a fixed exchange rate regime the central bank can raise interest rates and bring about an increase in its foreign assets.
However, in
order to preserve the increase in reserves, the central bank has to prevent the credit expansion that would result from the exchange of local for foreign currency, by reducing its credit by the same amount.[6] If it does so, international reserves will increase and their expansionary pressure will have been prevented. This phenomenon should
be viewed as a domestic interest rate rise with minimal, if any, short-run consequences on
M0. It will, however, be an inducement to holders of foreign assets to shift their worldwide portfolio in favour of the assets
of the country which has raised its interest rate. D will rise but not at the expense of a decrease in M0.
Sterilisation does not imply a modified
stance of monetary policy, but only the achievement of a different portfolio
structure. Sterilisation by the central bank of foreign capital inflows entails a reduction of A in order to accommodate the increase in IR. Commercial banks will experience a reduction in A but also a rise in D held by non-residents by the same amount.
The domestic credit provided by the central bank, in this example simply A or (M0 – IR), will have shrunk. This reduction in central bank
credit to the commercial banks occurs simultaneously with a compensating increase in non-resident
bank deposits, so that total credit to
the economy is not diminished and no contraction in the amount
of currency in circulation
has taken place.
The conclusions derived above
are arithmetically incontrovertible, as they are accounting identities. But it is useful to spell them out in detail in
order to emphasise the relationship between the different variables considered, given that economists’ reports, politicians’ statements
and press articles reveal a serious confusion about them.
An
additional discussion of the fixed-exchange-rate special case (known in the literature as the Mundell-Fleming model)[7]
would reinforce the original conclusion reached with the simplified accounting model used above: the central bank cannot induce immediate or short-run contractions in the money supply. This conclusion in no way contradicts the possibility for the money supply to increase if, for example, under a fixed exchange rate the demand for money is growing because
nominal income is
rising. Nominal income
may increase as a result of economic
growth and perhaps because
of some imported world inflation, or as a consequence of the evanescent effects of past exchange rate depreciations. In such situations the central bank will automatically provide the increase in the demand for cash. The Bank will finance these increases from its own credit or from increases in international reserves, depending on the circumstances.
Furthermore, irrespective
of the exchange rate regime, it should be emphasised that for money supply reductions to take place in the short run, unrealistic portfolio shifts
by the public are required away from its daily
cash needed for transactions into non-liquid, interest-bearing deposits. A symmetrical treatment would lead us to similar
conclusions when dealing with money supply
expansions.
It
could be argued
that the discussion so far has ignored the price
level and that if prices fall
when the central
bank reduces the nominal
supply of money, the real purchasing power of cash balances will be maintained. Large short-run price swings,
however, can be ignored because,
as will be argued, price movements reflect the delayed
outcome of monetary actions in the distant
past. Thus, while monetary policy tends to work as predicted in theory, i.e. inflation is a monetary phenomenon
caused by excessive central bank credit expansion,
contemporary price rises and money movements are the result of shocks, monetary or otherwise, that occurred months and sometimes even years earlier. This means that the monetary expansion witnessed
at any moment is the outcome of past decisions taken by the central bank or by the government which have given rise to a delayed increase in the demand
for money.
The reasoning and conclusions above can also be applied to a system in
which banks hold settlement balances
at the central bank either voluntarily or because of
reserve requirements. If
banks have cash balances at the central
bank, they will be able to reduce them if the central bank contracts M0. They will actually be forced to do so in order to
fulfil their payment obligations. This withdrawal is equivalent to central bank credit or to rediscounting of government paper.
[5] This conclusion
holds even in systems where banks are required to hold reserves at the central bank. In those systems commercial banks’
settlement balances have to be added to currency in circulation to calculate base money. Under these circumstances the central bank can increase or decrease the amount of base money since its immediate variations will come from the holdings
of base money of commercial banks and not from currency.
The transmission channels of monetary policy
If
the nominal money supply is neither a lever available to induce immediate changes in the stock of currency nor a trigger
for short-run movements in the general
price level, how does monetary policy
work?
It
was shown above how fixed exchange rate allows the central bank to
vary the level of international
reserves but not the monetary base. The reason is that under a fixed exchange
rate the supply of money is demand determined even in the medium
or long term. To induce
changes in the demand for money, a flexible
exchange rate is required. Under such an arrangement a monetary expansion can be brought about through the various channels that influence
the nominal exchange
rate. One is for the central bank to engage in non-sterilised purchases of foreign currency. In the central bank’s balance sheet this will be reflected by parallel and equal movements in IR and in M0: as it purchases foreign currency, the central bank will induce an increase in the assets held by commercial banks on their central bank accounts.
In our simplified accounts, commercial banks will hold a greater amount
of M0 (non-interest-bearing deposits) at the central bank. As they withdraw these non-interest-bearing assets and try to
place them among borrowers, lending rates will fall and banks will also be
compelled to lower the interest rates
paid to depositors. However, since in the simplified balance-sheet presentation the government is the only borrower from the banks, the excess non-interest-bearing cash in the hands of the banks will induce them to reduce the interest rate paid to
depositors as well.
As
interest rates fall people
will either spend more on goods and services, thus putting pressure on the current account
of the balance of payments, or purchase foreign assets. These outflows will tend to depreciate
the exchange rate, since one cannot presume that a home-made credit expansion will automatically bring in compensatory foreign flows to match the additional aggregate demand and to finance the resulting balance-of-payments deficit.
The depreciated exchange
rate will filter through to the price level via various channels. One is direct: the local currency cost of imports goes up, as do the prices
of exportable goods sold domestically. These rises will encourage increases in the prices
of local goods which are a substitute for or complementary to foreign ones.
When imported inputs and capital goods become more expensive, firms will be induced to raise prices to maintain
profit levels and to generate a flow of income sufficient to replace
their capital goods.
All these price
rises will, in turn, increase the nominal demand for cash. Since depreciation
shocks in Mexico have invariably been associated with a fall in internal demand, protracted price adjustments may also reflect the restoration of profit margins made possible by the gradual recovery
of internal demand to its former levels.
How the increased demand for cash produced by higher inflation manifests itself and is then met is key to the transmission mechanism. Some of the additional currency is supplied as the demand for it rises in response to the immediate
price increases. But the price adjustments triggered by a depreciation
of the exchange rate are spread over time, partly because expectations take time to adjust,
and partly because non-traded services contracts
are frequently staggered so that prices may take several years to catch up (Gil Díaz and Carstens (1996a and b)). As price rises
spread, the demand for nominal currency undergoes futher increases. This demand
shows up as currency is withdrawn from the banks through the exchange of D for M0.
As
noted above, the daily lubrication of the payments
system traditionally
performed by central
banks ensures that the additional demand for money is properly taken care of. As banks lose deposits while providing currency to the public, they will turn to the central
bank to replenish their shrinking liabilities with central bank credit for an equivalent amount. The central bank injects or withdraws credit every day in response to such fluctuations: it accommodates the demand for currency. In
its endeavour to ensure that the payments system clears every day, the central bank automatically supplies
the cash which the public demands. Thus, credit expansion
fuels inflation through its effect on the exchange rate and the ensuing inflation, in turn, induces currency expansion.
The two-way mechanism described
above is helpful
to illustrate the linkage between policy actions and subsequent
events, but is not likely to occur since it is uncharacteristic for central banks to cynically
purchase foreign currency in order to inflate. More
realistic scenarios must be looked for.
One is to have the central
bank determine interest rates.[8] Most central banks implement daily
liquidity expansions or contractions by charging or paying an overnight or very short-run interest rate.
This rate will influence the whole structure of interest rates in a variety of ways. Since the overnight rate is viewed by the market
as a benchmark, arbitrage possibilities will ensure that market
interest rates on some 28-day instruments will tend to equal (1+d)28, where d is the daily expected central bank interest rate. In turn, other term deposits
will react to changes in the one-month
rate so that the one-day
rate will end up influencing the overall term structure of interest rates.
Another
powerful transmission mechanism will be the implicit announcement effects of an interest rate change. An adjustment of the daily interest rate by the central bank may signal
to the market that the central bank aims for a tighter
or looser stance. Such a
change could have an amplified effect on the level of interest rates beyond the impact via the arbitrage mechanism described above,
if the markets believe that the change portends a trend, the continuation of a trend and even, if necessary, a future sharpening of the stance.
If,
as argued above, the very short-run interest elasticity of the demand for currency is zero, what purpose
can be served by trying to raise or lower interest rates? If a rise in the interest rate will have insignificant effects on the public’s preferences between currency and deposits, the central bank would
appear to be severely constrained in trying to influence the amount of currency in circulation.
But we shall see that it can do so, albeit with a significant delay, through its power to influence the determinants of the demand for currency.
In
the short run the central bank has virtually
no possibility of altering the demand for currency and, consequently, its supply, but its ability
to change interest rates has powerful, if delayed,
effects on both variables. A
rise in interest rates, for instance,
will depress consumption and investment, resulting
in a dampening of aggregate demand.
As this happens, the general price level will fall or rise less than it would otherwise have done.
The drop in inflation will make the nominal demand for currency lower
than it would have been without the change in the policy position of the central bank.[9]
Thus, the eventual
fall in the supply of money, or its smaller increase, is brought about not by directly reducing
its nominal supply but by
influencing the economic
determinants of its demand.
A
higher interest rate will produce other effects that reinforce those already described. It will attract
capital inflows and, through its negative impact on aggregate demand, will tend to reduce the demand for foreign currency or increase its supply through its effect on some current account transactions of the balance of payments. Both outcomes will tend to appreciate the exchange rate and, thereby, contribute to reducing inflation. In some countries
the direct link between the exchange rate and prices will be immediate and strong. In others,
the relationship between interest rates and aggregate demand may dominate.
It
may also be of interest to consider the opposite case of a rise in inflation. If the central
bank lowers its interest rate, the end-result may be
to stimulate aggregate demand and depress the exchange rate. Both developments will tend to raise inflation. Higher inflation will increase the nominal demand for currency, and the daily lubrication of the payments mechanism by the central
bank will translate this higher demand into a correspondingly higher degree of monetary accommodation.
An
expansionary stance
of the central bank therefore will gradually filter through to the price level and with it to the nominal demand for currency because of the time required to go through the different steps in the transmission mechanism and, as argued before, because
of the fact that contracts are typically
staggered over time. Inflationary expectations play an important part in determining the duration of contracts governing wages, rents, tuition,
etc. To the extent that contracts
are not revised every day and overlap over time, price adjustments may take years to work through, whether
they originate in shifts in aggregate demand or changes in the exchange
rate.
In
this fashion, the observed contemporary price rises are the outcome of past modifications in the central
bank’s policy stance,
or of previous supply
shocks or political disturbances. It is in this sense that, following Friedman (1968),
monetary policy
has delayed and changing effects on prices.
[8] The central bank can do this through various mechanisms: it can use keynote tender operations at fixed rates
(fixed rate tenders), or it can target short-run
or overnight interest rates through variations
in its provision of liquidity.
[9] The total effect of the policy change on the demand for money is somewhat more complicated. Sooner or later lower inflation will bring about a decrease in nominal interest rates. Therefore, the decrease in the demand for money due to lower inflation will be compensated, to some degree, by an increase in its quantity
demanded. But from experience, it can be safely concluded that the first effect will eventually
dominate and lower inflation will generally mean a lower nominal demand for currency than would otherwise have occurred
Mexico’s monetary policy
While the transmission channels
examined above differ in intensity across countries
because a vast variety of institutional arrangements and historical experiences exist, they represent the range of options available to
central bankers and policy-makers around the world. Depending
on preferences, on the degree of central
bank independence and credibility, on
the level of present and past inflation, on the size and openness
of the economy, among other variables, the central bank will choose a set of objectives and operational variables
to carry out its task.
Some countries, such as Canada, have chosen a floating exchange
rate, with foreign exchange interventions implemented in such a mechanical and predictable fashion
that their regime can be likened
to a free float. The central bank modifies
its stance if it feels that circumstances call for such a change. Others,
such as New Zealand, have also maintained a free float, albeit without lubricating interventions in the foreign exchange market. In the case of New Zealand, the central bank favours issuing statements that send a signal to the market and exert a subtle and sophisticated influence on freely determined market interest rates (see Reserve Bank of New Zealand (1992), p. 73).[10]
The reference to New Zealand seems
appropriate because
recently the course of monetary policy in Mexico has shown some parallels
with that country’s experience, although Mexico’s evolved independently.
From the onset of the 1994 peso
devaluation and the ensuing financial crisis, to the present day, the Banco de México has continued its practice of
setting its settlement cash target daily to accommodate the demand for currency, but it has radically modified
its mode of intervention.
Before the crisis fixed rate tenders were often
used to set each day the interest rates at
which it would conduct its intervention
to withdraw or inject liquidity.
The
crisis, however, brought about a nominal depreciation
of the exchange rate of more than 100 %, high and variable inflation and, initially, a
severe slump in the real economy. The situation was further complicated by highly volatile
perceptions abroad of Mexico’s prospects.
In
such a context it would have been highly inadvisable to attempt to set an interest rate for central bank transactions. Too low a rate,
perhaps causing real interest rates
to turn negative, would
have encouraged lending and higher inflation. Too high a level would have aggravated the problems faced by borrowers and the portfolio
difficulties experienced by
commercial banks. Thus, it was decided
that the accommodating stance of the central
bank as regards the demand
for currency would
be complemented with a freely-floating, market-determined interest rate on the central bank’s daily
operations with commercial banks.[11]
Given this arrangement, the question arises
as to the type of procedure the central bank could use if it considers that intervention
is appropriate. A variety of circumstances could call for intervention:
for example, the central bank might
believe that inflation is rising too quickly; it could fear that rapid exchange
rate movements might provoke a bandwagon effect; or it could
view the rate of growth in total credit as excessive.
To
influence monetary conditions, the Reserve Bank of New Zealand has relied on carefully
prepared statements of its Governor or senior officials. The Bank for its part uses a daily
announcement of its target for the cumulative or average
amount of settlement balances for signalling purposes. Movements in the target for the cumulative balance
are translated into changes in the amount of settlement
balances that the Bank injects or withdraws at auction-determined interest rates in its daily operations
with commercial banks. Movements
in the cumulative target have been so minuscule that they amount to something
akin to the moral suasion that many central banks have sought to impose through policy statements.[12]
With regard to current monetary management in Mexico, the daily operations of liquidity injection
or withdrawal are effected in the context of
a zero reserve requirement for the accounts
of commercial banks at the central bank. Given the characteristics of the Mexican payments system,[13]
banks are confident that they will be able to overdraw or add to
their central bank accounts if on any given day there is insufficient or
excess settlement cash. In practice, imbalances are relatively
small because of the daily targeting procedures and the close coordination between the Treasury and the Bank. The Treasury will not issue a payment on t of which it did not give notice on t-1, and commercial banks, through which all tax revenues
(including customs duties) are channelled, give notice on t–1 of Treasury income to be deposited
on the Treasury’s account at the central
bank on t. With daily settlement cash targets always
including compensation for t–1 errors, the amount of net settlement balances demanded by commercial banks on any given day is virtually nil. Complementing these procedures is a central
bank policy of aiming for a zero, or near-zero, excess supply of settlement cash on any given day.
A
crucial element of the zero average reserve requirement scheme is the daily announcement by the Bank of its target for the cumulative
balance of commercial banks’
accounts at the central bank for the next day’s market opening. Thus,
for example, announcing a zero objective for the cumulative balance
implies a neutral monetary policy,
while a movement to a negative
target would signal a tightening of monetary conditions.
Overdrafts can be easily induced.
The Bank injects or withdraws credit, as the case may be, in its daily operations in order to accommodate the fluctuations in the demand for currency. If less credit tan needed
is injected, the banking sector as a whole will incur an overdraft on
its central bank accounts.
Above conclusions do not hinge on a system
of zero reserve requirements,
nor on a banking system
that holds practically no settlement balances, nor on the resulting
identification of currency with the monetary base. If banks were required to hold a fraction of their liabilities deposited at the central
bank, an error in targeting the daily cash needs
(e.g. a shortfall)
by the central bank would be cushioned
by withdrawals from their accounts. This mechanism does not differ in any essential
way from the workings of a system based on zero reserve requirements. The same applies to the process,
favoured in some countries, of rediscounting of government paper by the commercial banks at the central bank. Under a
system of zero reserve requirements the central bank can also set a daily cash target different from zero, or alter it if it wishes to influence monetary conditions, with effects that are indistinguishable, analytically and quantitatively, from what would occur under a system in which banks need to hold settlement
balances or are required to maintain
reserves at the central bank.
Graphs 1 and 2 illustrate the development of key monetary variables and their relationship with policy changes.
The graphs show the behaviour of
the spot exchange
rate, of a leading market interest rate and of the changes in the Bank’s
target for settlement cash throughout
1995 and 1996. The interpretation of the latter is as follows: from 2nd to
23rd January 1996, the central
bank merely accommodated the daily demand for currency, that is, it left at zero the amount of commercial banks’ settlement balances at the central bank.
On the latter date it changed its stance from neutral to – 5 million pesos and on 25th January to –
20 million, where it was held until 7th June, when it was changed to –
30 million.
A
10 million peso shortage induced
by the central bank until the closing 28th day of the cycle,
meant that some commercial banks ended up
paying twice the Treasury bill rate on the amount
of their overdrafts at the end of the 28-day averaging period used to calculate their
compliance with the reserve requirement. The reserve requirement,
or the amount that banks have to hold on average
in their accounts
at the central bank, is
at least zero.
The amounts involved mean nothing quantitatively: 10 million
pesos amount to 0.0000063 of the average
liabilities of the banking system
in 1996. The interest cost involved is also insignificant. Yet, as the graphs show, these small movements
in the central bank’s daily accommodation were signals that the market took into account and to which they reacted accordingly.
Graph 1
Banks’ funding rate, the spot exchange
rate and the objective
for accumulated balances, 1995*
*
Accumulated settlement balances
in millions of pesos are shown on the top scale.
[10] Signalling is a technique
widely used by central banks (Borio (1997)). Regarding the supply of bank reserves
by the Bank of Canada,
Borio reports (p. 26): “Supplying, say, a somewhat larger
amount than that targeted by banks is expected
to put downward pressure on the overnight rate. It is still an open question, however, how much of the downward pressure occurs through a mechanical liquidity effect or,
more fundamentally, through the
signal conveyed regarding monetary
intentions.”
[11] This strategy is not unique.
Again quoting Borio ((1997, p. 50): “The initial move towards more market-oriented means of policy implementation away from standing facilities and, in some countries, the greater focus on quantitative objectives for operating and/or intermediate
aggre- gates went naturally hand in hand with implementation strategies where central banks gave less guidance about desired interest rates. At a time when reducing inflation
was paramount, these policies were also seen as a way of shielding
central banks from social and political
resistance to unpalatable increases in interest rates”.
[12] See Tait and Reddell
(1992, pp. 72 – 73). As to the question “How can mere announcements have such a critical effect?” Borio (1997, p. 57) notes: “The answer perhaps lies in the fact that as a monopolist supplier of settlement balances, the central
bank could, if it so wanted, set the overnight rate.”
[13] Daily government operations
are known with certainty one day prior to
their taking place. Up to mid-1995 all the daily movements in the government and bank accounts
at the central bank were offset daily through a second auction of settlement
cash known as the ”milkman’s market”,
as it was conducted before banks open for business, at the previous closing- day value date. This system was transformed into the present one, which allows banks to draw settlement balances
within the 28-day
zero reserve requirement period.
Within this period the market knows that the central bank will offset completely all movements on its accounts
by means of a second daily auction of settlement
balances which is conducted at least each closing 28th day of the cycle. The rule of fully offsetting
account balances is altered only when the central bank deliberately modifies its stance,
thereby forcing the banks to overdraw their current accounts at the central bank. But as noted in the text, these overdrawings
are so small, that they can only be interpreted as signals: in 1995 the largest overdrawing
amounted to 40 million pesos, or barely $5 million out of $200 billion of bank liabilities.
Graph 2
Banks’ funding rate, the spot exchange
rate and the objective
for accumulated balances, 1996
*
Accumulated settlement balances in millions of pesos are shown on the top scale.
A few of the several instances in which the central bank felt it appropriate to act could be analysed. On 7th June 1996 the exchange rate started to depreciate too quickly. This led the Bank to withdraw 30 million pesos from the market instead of the previous 20 million. Shortly afterwards the exchange rate again showed some nervousness and the amount was raised to – 40 million. Graph 2 illustrates how interest rates reacted immediately. It also shows that periods of exchange rate appreciation were associated with falling interest rates, as in the period from 22nd February to 14th May 1996. An example of the opposite movement is the period from 3rd to 29th July 1996. In both periods shifts in the exchange rate and in interest rates occurred in the absence of policy changes, simply as the result of natural market forces reflecting how, when the exchange rate depreciates, the demand for funds to purchase foreign exchange leads to higher interest rates, given that the neutrality of the central bank’s stance implies that it will not inject excess funds into the system.
Another interesting feature of the variables depicted in the graphs, is that the announcement effect of shifts in the central bank’s stance seems to fade rapidly. A negative
(restrictive) stance maintained long enough can be
found to be associated with falling interest rates and vice versa. Possibly because of the very small amounts involved, the changes in the stance more than its sign or absolute
value therefore seem to be what matters.
The transmission channels of the Mexican
economy have been well documented in several research papers, some of which are quoted in the list of references below. These channels are closely related to the degree of
openness of the economy. Prices in Mexico have a long history of sensivity to the exchange
rate. This relationship and its speed have been reinforced by the broad opening-up to trade of the Mexican
economy since 1985: current account transactions, excluding interest flows, accounted for 34 % of GDP in 1993, 38 % in 1994, 58 % in 1995 and 60 % in 1996 (national accounts figures).
A
recent study (Pérez-López (1996)) has shown
that in Mexico the behaviour of the exchange
rate and that of wages maintain a close
relationship with the evolution of prices. Moreover, this study suggests that a 10 % rise in wages causes a 6 % increase in the price level over a seven-month period, while a 10 % depreciation
of the exchange rate generates a 4.5 % increase in the price level during the subsequent eight months.
The cause-and-effect relationships are more complex than what statistical correlations appear to corroborate.
If the nominal exchange rate had remained fixed from 1994 until the present, for instance,
most analysts would
agree that nominal wages would have exhibited far smaller increases, so that,
even though salary increases bear on prices,
wages are also influenced by exchange rate depreciations. Nominal wages inevitably adjust upwards after a devaluation in response
to the price increases provoked by the depreciated exchange
rate, and wage hikes will be reflected in price jumps that, in turn, will feed into wages again until, if the nominal exchange rate remains
at its new, depreciated level,
these ever-smaller adjustments peter out. Causation thus appears to be from devaluation to prices, with wages playing
catch-up and nudging
along price increases with another ultimate cause.
Economic causality appears to be from interest rates to the nominal exchange rate and then to output,
and from exchange rate movements to prices, from prices to wages and again from wages to prices. The output relationship has been left until the end because in recent years the impact of
exchange rate movements on variations in real output in Mexico has been rather modest under conditions of high inflation. With fast inflation nominal exchange rate variations swamp real ones. Of course, from a welfare point of view, the output link is more important. Again,
it has been the Mexican experience (documented in Pérez-López, (1995)) that deviations of the real exchange
rate from its stationary level create fluctuations
in output. A depreciation
of the real exchange
rate is associated with a fall in output and vice versa for an appreciation. Since shifts in monetary policy will have effects on the nominal
exchange rate, the real exchange rate will be transitorily affected by modifications in the stance of the central bank. The effect of the real exchange rate on output will of course also feed into changes in the demand for money.
The graphs contain another
lesson. A neutral
accommodating stance is not equivalent to inaction on the part of the central bank. Free floating of interest rates and the exchange rate implies that both variables
fluctuate in a sychronised way to absorb shocks. If capital flows out of the country the adjustment will be spread between both variables, credit will become
scarcer and the exchange
rate more depreciated.
An invariant supply
of liquidity by the central bank,
understood as modifying its credit solely to accommodate the daily fluctuations in the demand
for cash, implies
that speculators cannot count on the provision of central bank liquidity to finance their outflows. Therefore, interest rates rise as capital
flows out, making speculation more expensive and helping to stem or halt the outflow. In a Canadian context,
this would be equivalent to monetary conditions showing compensating movements with no intervention by the central bank. Therefore, a stance of merely accommodating the demand for currency is indeed a policy, viz. a policy consisting of not allowing conditions to loosen when there is a run. A symmetrical reasoning applies to
ex ante capital inflows.
Annex 1
The operational framework of the Banco
de México
The operational framework adopted
by the Banco de México
to implement its monetary policy comprises a reserve requirement with averaging around a level of zero reserves over a 28-calendar-day maintenance period. The regime was introduced in March 1995 to allow the Bank to send
quantitative signals
to the money market
without determining interest rate levels.
Under the scheme, the Bank does
not remunerate positive settlement balances nor does it charge for overdrafts posted
at the end of each day in
the commercial banks’ current accounts at the central
bank. However, it
charges a penalty rate at the end of the maintenance period, if the cumulative
(average) balance is negative. The cumulative balance is defined as the sum of the daily positive
and negative (overdraft) settlement balances. The penalty is meant
to give banks an incentive
to end the maintenance period with a zero cumulative balance and is equal to twice a market-determined rate. The magnitude of the penalty
is set at such a high level in order to make the net cost of end-of-period negative cumulative balances similar to the net cost of holding end-of-period positive cumulative balances.[14]
In
order to prevent large fluctuations in commercial banks’ account balances and to reduce the capacity of banks to take leveraged
positions to influence interest rates, limits were established for the overdraft facility and for the amount
of positive balances
that count towards the cumulative balance (Graph 3). The limits were set on an individual basis related to
each bank’s capital.
The Bank intervenes every day in the money market injecting or with- drawing liquidity. The Bank determines the amount of its intervention so that the sum of all banks’
cumulative balances at the end of the day adds up
to a predetermined
amount. The Bank announces every day the predetermined amount
and uses it to signal
its monetary policy intentions
[14] The net cost of an end-of-period negative
cumulative balance is equal to twice a market- determined rate minus the return from investing at market rates the funds obtained through the overdraft. The net cost of an end-of-period positive cumulative balance
is equal to the return forgone by not investing
the funds in the market.
Graph 3
Positive and negative balances
in banks’ accounts
(Table 1). Thus, a reduction
in the Bank’s target for the cumulative
balance would indicate its intention to tighten its monetary stance, whereas an increase would signal its intention to ease policy.
Table 1
The Bank does not use an official
interest rate nor does it set maximum or minimum levels
for interest rates during
its interventions
in the money market.
All interventions
are carried out through auctions (Table 2) in which the Bank determines the quantities
and the market freely sets the interest rates.
Table 2
The Bank intervenes every business day to bring the cumulative balance to the announced target. In doing so, it takes into account
all transactions that have an impact on the balance in the banks’ settlement accounts at the central bank, such as changes in currency demand, government receipts
and disbursements, foreign exchange
interventions and the falling-due of previous open market
operations. The Bank has complete prior information
on all these operations, except
for cash deposits or withdrawals made by credit institutions to meet changes
in the demand for currency. The Bank credits (or debits)
banks’ current accounts on the same day as banks deposit
the notes taken
from the public or withdraw them. Therefore, each
day the Bank includes its own daily forecast of changes in the demand
for notes and coins[15] in
its estimate of the intervention
in the money market (Graph 4 and Table 3).
Graph 4
Demand for notes and coins
Table 3
Table 4
[15] In Mexico,
commercial banks’
demand for settlement balances is relatively small,
making the monetary base almost equal to the demand for notes and coins. This is due to the following reasons: (a) the Bank offsets all changes in the settlement balances of the commercial banks’ accounts; (b) it provides daylight
credit; (c) commercial banks can overdraw their accounts
at the end of the day during the maintenance period; and (d) there is a pre-settlement market (see Table 4) in which commercial banks are allowed to borrow and lend among themselves
after their settlement positions are known
and third parties are no longer permitted to transact.
Annex 2
Interest rate impact
of selected shocks:
econometric evidence
To
assess the effect on short-term interest rates of changes in the Banco de
México´s target for the cumulative balance,
a regression was run using the interbank overnight interest rate as the dependent
variable and the 30-year US bond yield, the peso/dollar exchange rate, liquidity
shocks[16] and the Bank’s target for the cumulative balance (CB) as explanatory vari- ables. The results obtained show that short-term interest rates react to the signals sent by the Bank but not to transitory (unintended) liquidity shocks.[17]
The sample period runs from early September 1995, when the Bank established
the system of zero average reserve requirements with synchronised maintenance periods for all banks,[18] to
end-December 1996, one month after it changed its monetary stance for the last time.[19] Daily data were used.
To
provide a test of the stability
of the coefficients obtained, the sample was divided in two. The first sub-period
runs from 7th September 1995 to 14th March 1996, when the Bank imposed
limits on the settlement balances counting
towards the cumulative balance. The second subperiod runs from 15th March 1996 to 31st December
1996. The results for the whole period
and for the two sub-periods are shown in Table 5.
Table 5
Interest rate impact of selected
shocks
[16] The source of the liquidity
shock is a central bank error in its daily forecast for notes and coins in circulation. A negative shock
occurs when the actual demand
for notes and coins exceeds the demand forecast by the Bank and thus commercial banks end up overdrawing their accounts at the central
bank to satisfy the public’s
demand for notes and coins.
[18] From March to August 1995 the maintenance period for reserve requirements of the various
banks was overlapping in time.
[19] The last modification of the policy
stance was on 8th November
1996 (this paper was
completed in August 1997).
References
Borio, Claudio
E.V. (1997): “Monetary policy
operating procedures in industrial countries”. BIS Working Papers, No. 40, March.
Copelman, M. and A.M. Werner (1996): “Exchange rate policy and the
bank lending channel:
Sectoral evidence from Mexico”. Working Paper Series, Federal Reserve Board, Washington, D.C.
Fleming, J. Marcus (1962):
“Domestic financial policies
under fixed and under floating exchange
rates”. IMF Staff Papers, 9(3), pp. 369 – 80.
Friedman, Milton (1968):
“The role of monetary policy”.
American Economic Review, 58(1), pp. 1–17.
Gíl Díaz,
Francisco and Agustín
Carstens (1996a):
“One year of solitude: some pilgrim tales about Mexico’s
1994 –1995 crisis”.
American Economic Review, 86(2), pp. 164 – 69.
Gíl Díaz, Francisco and Agustín Carstens (1996b): “Pride and prejudice:
the economics profession and Mexico’s financial
crisis, 1994 – 95”. Document prepared by the Dirección
General de Investigación Económica, Banco de México.
Goodhart, Charles (1987): “The monetary base”, in John Eatwell, Murray Milgate and Peter
Newman (eds.): The
New Palgrave dictionary of economics. London: Macmillan.
Juan-Ramón
Hugo (1996): “The daily conduct of monetary policy in Mexico”. International
Monetary Fund.
Mundell, Robert, (1963): “Capital
mobility and stabilization policy under fixed and flexible exchange
rates”. Canadian Journal of Economics and Political Science, 29(4), pp. 475 – 85.
Pérez-López, Alejandro (1995): “Un modelo
de cointegración para pronosticar
el PIB de México”. Banco de México,
Documentos de Investigación, No. 9504.
Pérez-López, Alejandro (1996): “Un estudio
econométrico sobre la inflación
en México”. Banco de México, Documentos de Investigación, No. 9604.
Tait John and
Michael Reddell (1992): “The operation of monetary
policy”, in Reserve Bank of New Zealand:
Monetary policy and the New Zealand financial
system. Third edition.
Thorne Alfredo (1997): “Assessing the effectiveness of monetary policy”, in JP Morgan: “México’s monetary
policy”. Economics Research Note,
México, 24th January.
