1

KINGSTON UNIVERSITY, LONDON

SCHOOL OF ECONOMICS

Monetary Economics in Developing Countries

(FE3178), 2009-2010

Monetary policy transmission

 2

Monetary policy transmission

Introduction

We discussed features of monetary policy institutions in

developing countries and how these features affect

economic performance.

Today we try to answer the following question:

How do a central bank’s actions work their way through

an economy?
 3

The transmission mechanism of monetary policy

Understanding how monetary policy actions are actually

transmitted to the rest of the economy is an ongoing

subject of study for economists.

The Symposia on the Monetary Transmission

Mechanism in the Journal of Economic Perspectives,

1995, Volume 9, Number 4, pages 3-96, contains an

interesting set of papers on this topic.

Even though institutional features will likely determine

the exact transmission mechanism for an individual

economy, several hypotheses have been put forward with

the aim of explaining this process.

 4

All these frameworks assume that the monetary policy

transmission process begins with a change in the

monetary policy stance of the authorities.

An example would be a decrease in the money supply

originating from a policy aiming at altering bank reserves

or changing a short-term interest rate -e.g. the interbank

interest rate, which is the rate commercial banks charge

for lending funds to each other.

One method of transmission would be the workhorse

mechanism which states that a contractionary

monetary policy will affect output through an increase

in the interest rate, with this increase leading in turn to a

reduction in investment and output.

 5

Another possible transmission mechanism operates

through the exchange rate’s impact on net exports.

A reduction in the money supply will generate an

increase in the interest rate, which will subsequently

cause the exchange rate to appreciate, reducing the

amount of net exports and therefore aggregate demand.

This outcome, however, is contingent upon the fulfilment

of the ‘Marshall-Lerner condition’ which states that a

depreciation’s ultimate impact on the trade balance

depends on imports’ and exports’ exchange-rate-

elasticity.

A third possible transmission mechanism is embodied in

the bank lending or credit view.

 6

Banks here play a direct role in the transmission

mechanism of monetary policy: a lower money supply

decreases bank deposits, which in turn lowers bank

loans, and that works by reducing investment and output.

The transmission mechanism of monetary policy in

developing countries

Analysing monetary policy’s transmission mechanism in

developing countries traditionally focuses on measuring

financial liberalisation’s impact on credit availability.

Recall that the McKinnon (1973) and Shaw (1973) and

the Structuralist views (e.g. Taylor, 1983; van

Wijnbergen, 1982, 1983) hold opposing views on

precisely this question, as explained in detail in Chapter

4 (see also Auerbach and Siddiqui’s, 2004, survey).

 7

McKinnon and Shaw argue that higher interest rates

resulting from financial liberalisation raise output and

lower inflation in the short-run: financial liberalisation

leads to increasing savings and financial

intermediation, which boosts bank loans and private

investment in the medium term.

A key assumption underlying this approach is that

liberalising interest rates works by mobilising idle

resources to productive use.

Note that the degree of substitutability between bank

deposits and resources in the informal credit market is

crucial.
 8

In contrast, the structuralists argue that financial

liberalisation polices may increase the marginal cost of

funds in the ‘unorganised money market’ by taking

resources away from it (van Wijnbergen, 1982; Taylor,

1983), with those unable to access the formal loans

market suffering from the contraction of funds.

Thus the structuralists predict that a monetary policy

tightening, i.e. reduced credit availability, and

corresponding higher interest rates and costs of financing

working capital, which leads to contracting output and to

an increasing rather than to a decreasing inflation rate.

This effect is equivalent to an adverse supply shock

leading to stagflation.
 9

A further prediction from this approach is that, if

effective in reducing real aggregate demand, a

contractionary monetary policy works by reducing

investment and not consumption, with negative

consequences for an economy’s growth prospects.

For the sake of clarity, we summarise the McKinnon and

Shaw and the structuralist transmission mechanisms as

follows

McKinnon-Shaw

FL ⇒ i ↑⇒ s ↑⇒ BL ↑⇒ I ↑⇒ Y ↑

Structuralists

FL ⇒ i ↑⇒ ILM ↓⇒ MCF ↑⇒ I ↓⇒ Y ↓
.
 10

In the above mechanisms ↑ indicates an increase, and ↓ a

decrease in a given variable.

The variables are defined as follows:

i is the real interest rate, I investment, Y real output, E the

nominal exchange rate (an increase is a depreciation), BL

bank loans, FL financial liberalisation, s are private

savings, ILM informal loans market (or unorganised

money market), and MCF stands for the marginal cost of

funds.
 11

A benchmark model

Montiel (1991) develops an analytical model considering

salient features of a developing economy’s monetary

dynamics.

In particular, he attempts to explain how a central bank’s

actions are transmitted to the rest of the economy.

He argues (Montiel, 1991, p. 83) that “Curiously,…in

spite of the prominence given to monetary policy in the

developing economy setting, the transmission

mechanism for monetary policy in a typical developing

country has not been studied extensively and

consequently is not well understood”.

 12

Even though more empirical evidence on the subject now

exists, Montiel’s analytical framework is still relevant.

The model in question brings into play several features

like domestic currency, bank deposits, foreign currency,

land, and physical capital.

The model also incorporates a curb market –an important

feature in an undeveloped financial system.

Thus, for instance, economic agents can acquire assets

such as land and physical capital either by obtaining

liquidity through the curb market or from the financial

system.

Interest and exchange rate determination also feature in

this model. In developing economies, interest rates are

 13

often capped by formal regulations, therefore limiting the

role of financial intermediaries in linking savers and

borrowers; exchange rates are freely determined in a

parallel or black market, whereas the official price is

fixed.

This multiple exchange rate market setting arises chiefly

due to the monetary authorities’ desire to isolate critical

domestic prices, like oil, from external fluctuations (See

Calvo and Reinhart, 2002).

In the setting described so far, traditional monetary

policy instruments, such as interest rates, that are crucial

in developed economies (e.g. Bernanke and Blinder,

1992) are not as important in central bank policymaking

in developing countries. Instead, instruments such as

central bank domestic credit, legal reserve requirements

 14

for the banking system, and exchange rate market

interventions play a prominent role in developing

countries.

Accordingly, Montiel explicitly models households, the

government, the monetary authorities, and the banking

system in a Mundell-Fleming framework.

Important assumptions are as follows: full-employment

holds continuously and changes in aggregate demand

only lead to changes in prices; agents form expectations

rationally, this assumption implying that expectations

about future developments in exchange rates and in

prices induced by current policies are allowed to affect

current macroeconomic conditions.

Montiel’s framework leads to several conclusions.

 15

Changes in monetary policy impact the effective degree

of financial repression by affecting the fashion in which

that policy taxes households, and by altering the structure

of households’ portfolios as a response to financial

repression.

The model also produces general equilibrium effects that

work via the impact of policy changes on the parallel

foreign exchange market premium, the economy’s stock

of foreign assets, and expected inflation.

However, the effects from changes in the stock of foreign

assets (that are a by-product of policy changes) are only

observed with a lag.

 16

Additionally, variations in central bank credit to the

banking system and reserve requirements work their way

through the economy via their impact on wealth, the loan

interest rate, and fiscal effects. Raising administered

interest rates is contractionary.

Finally, the impact of a foreign exchange sale (akin to a

contraction in the money supply) is contractionary. So a

purchase of foreign exchange (an increase in the money

supply) is expansionary.

Carpenter (1999) further studies monetary policy’s

transmission mechanism in developing countries.

He develops an IS-LM model incorporating a regulated

banking system and an informal credit market.

 17

Importantly, Carpenter assumes that there are no bond or

equity markets.

The model’s main prediction is that increases in

monetary policy’s control instrument (the money supply)

are expansionary via the credit view, by which more

credit leads to more output.

This result is line with Montiel’s (1991) predictions, but

not with those of van Wijnbergen (1982).

Notably, the latter predicts stagflationary developments

following a monetary expansion.

The next section discusses the empirical evidence on the

topic.
 18

Empirical results concerning the transmission

mechanism of monetary policy in developing
countries

Measuring monetary policy’s impact on real and nominal

economic variables has been approached from various

perspectives

With the vector autoregression (VAR) method (to be

discussed below) popularised by Sims (1980) arguably

the tool macroeconomists employ most frequently when

dealing with such a task.

Sims argues that estimating large-scale

macroeconometric models reveals little useful

information, and demands ‘incredible assumptions’.

During the 1960s and 1970s evaluating the likely impact

of alternative policies was largely undertaken by

 19

constructing and estimating large-scale structural

macroeconometric models.

However, a pillar in this approach, i.e. assuming that a

model’s estimated parameters are invariant to the

relevant policy rule’s specification, was criticised in

Lucas’s (1976) influential paper.

But subsequently Lucas’s main critique regarding the

rationality of economic agents has been incorporated

into large scale structural econometric models (e.g. Fair,

1984), and macroeconometric models are still popular in

institutions such as central banks and other policymaking

institutions (see, for instance, Mankiw’s, 2006, remarks

on the topic).
 20

VARs have become a popular tool for the evaluation of

theoretical macroeconomic models in developed and

developing countries.

But various authors criticise the VAR approach and its

usefulness in examining monetary policy’s effect on the

economy.

We turn to reviewing several contributions to the

understanding of the monetary policy transmission

mechanism in developing countries, predominantly those

using the VAR approach.

As revealed in the analysis so far, tensions exist between

the somewhat conflicting predictions from various

analytical approaches.
 21

Using Korean data, van Wijnbergen (1982) finds

support for the structuralists’ approach. Particularly, he

shows that restrictive monetary policy leads to

stagflationary developments.

But that evidence is to some extent challenged

elsewhere.

For instance, Carpenter (1999) also analyses Korean

data, and discusses the relevance of credit provision by

the authorities and the existence of an informal credit

market.

His empirical modelling employs the VAR approach,

chiefly in estimating impulse response functions

depicting the trajectories of key macroeconomic

variables resulting from shocks to monetary policy

 22

instruments, in this case either M2 or the Bank of

Korea’s credit.

The VAR modelling considers Cholesky’s

decomposition as well as Bernanke’s (1996) more

structural identification strategy. Both methods yield

analogous outcomes.

Carpenter’s results are interesting.

He finds that, supporting Montiel’s (1991) and van

Wijnbergen’s (1982) models, there is a positive and

significant reaction in the price level following a shock

to the interest rate.

He argues that the reason for this response is that

increasing credit costs lead to higher input costs.

 23

Moreover, as predicted in Montiel’s model, he finds that

increases in money are expansionary.

However, Carpenter does not find evidence of

stagflationary developments found in van Wijnbergen’s

modelling.

He does find that contractions in money lead to a fall in

output, but also in prices.

Carpenter makes an attempt at rationalising his VAR-

based findings vis-à-vis van Wijnbergen’s predictions on

the path of the informal interest rate following a shock to

money or credit.

He actually finds that a positive policy shock leads to an

increase in the informal market’s interest rate; the

 24

rationale for this finding is that general equilibrium

effects work by increasing output.

That, in turn, leads to a higher demand for resources in

the informal sector, and so to a correspondingly higher

interest rate.

Other contributions to our understanding of monetary

policy’s transmission mechanism in developing countries

put more emphasis on interpreting shocks than to

reconciling them with relevant theories, as is the case

with VAR-based studies like Leiderman (1984) and

Kamas and Joyce (1993).

For instance, Leiderman estimates a significant and

positive association between money growth and inflation

in Colombia.
 25

Reinhart and Reinhart (1991) produce an interesting

analysis comparing the relevance of the neoclassical

synthesis and of business cycle theories in a developing

economy context.

This is a welcome effort, since some economists may be

inclined to rule-out ex-ante the neoclassical synthesis and

business cycle theories’ applicability in a developing

economy context.

Basically, the neoclassical synthesis postulates that

monetary policy can be effective in the short-run; due to

price stickiness, an unexpected monetary expansion will

increase output.
 26

But that effect will only survive until prices adjust.

(Goodfriend and King, 1997, explain developments in

this line of thinking.)

On the other hand, business cycle proponents do not take

into account the impact of monetary disturbances on the

economy (e.g. Nelson and Plosser, 1982).

Thus the findings which emerge from applying that

framework may only be able to account for a limited

amount of the many potential sources of output

fluctuations -chiefly technological ones. And that is

despite the vast theoretical and empirical literature

evaluating money’s impact on the real economy.

In contributing to this debate, Reinhart and Reinhart

(1991) key finding is that for Colombia a neoclassical-

 27

Keynesian story is more sensible than a business cycle

perspective in which money plays no role. That is,

money does affect output.

Importantly, they find that output falls following a

monetary contraction, so disinflationary policies are

costly.
 28

Figure 8.1
Standard monetary policy transmission mechanism

• Investment
Change in monetary policy stance
• Exchange rate and Output and
affecting bank reserves and/or short- → →
net exports prices
term interest rate
• Bank credit
 29

Table 8.1
Selected empirical studies on the transmission mechanism of monetary policy
in developing countries (in chronological order)
Econometric
Investigation Country Key finding
technique
Two stage least
van Wijnbergen (1982) South Korea Short-run stagflationary effect of restrictive monetary policy
squares (TSLS)
Colombia For Colombia money growth affects inflation. For Mexico two-way
Leiderman (1984) VARs
and Mexico causality between money and inflation
Reinhart and Reinhart Finds support for the neoclassical synthesis. That is, monetary shocks
Colombia VARs
(1991) affect output
Kamas and Joyce (1993) India and Domestic monetary policy does not affect output, but in both
VARs
Mexico economies output responds to changes in foreign money
Variations in domestic credit affect the balance of payments but not
the exchange rate. So it seems that Colombia’s crawling peg
Kamas (1995) Colombia VARs
effectively works as a fixed rather than a flexible exchange rate
regime.
Central bank credit significantly affects output, prices and interest
rates in the informal sector. Findings support Montiel’s (1991) model,
Carpenter (1999) South Korea VARs
but not the stagflationary response to monetary contractions in van
Wijnbergen (1982)
Transmission mechanism from financial institutions’ administered
Chong et al (2006) Singapore Time series rates is asymmetric across sectors in the economy. A monetary
tightening impacts the economy with a longer lag than an expansion
Note. – VAR: vector autoregression.