OF EXCHANGE · Flexible Price Monetary Approach ... IMF/World Bank formula chiefly in the areas of exchange rate adjustments, and reduction ... (BOP) measure. The use of this approach

A LONG-RUN MONETARY MODEL OF EXCHANGE RATE DETERMINATION FOR GHANA

Justin Tevie

Submitted in partial fulfillment of the requkements for the degree of Master of Development Economics

Dalhousie University HaMax, Nova Scotia

December, 1996

O Copyright by Justin Tevie, 1996

Nauonai uurïry of Canada

c~iuiiuu icqua I iauvi mi=

du Canada

Acquisitions and Acquisitions et Bibliographie SeMces services bibliographiques 395 Wellington Street 395, rue Wellington Ottawa ON K1A ON4 OttawaON K1AON4 Canada Canada

The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distriiute or sell reproduire, prêter, distri'buer ou copies of this thesis in microform, vendre des copies de cette thèse sous paper or electronic formats. la forme de microfiche/fïlm, de

reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni Ia thèse ni des extraits substantiels may be printed or othewise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

Dedication

To my parents

Table of Contents

Dedication List of Figures List of Tables Abstract Acknow1edgment

Chapters

1.0 Introduction

2.0 Adjustment Programmes and the Ghanaian Economy

2.1 Economic Adjustment Programmes Economic Recovery Programme Exchange Rate Policies

2.2 Financial Structure 23 Economic Review (1983 - 1994)

Monetary hdicators Exchange Rate Gross Domestic Product Balance of Payment, hports and Exports

2.4 Conclusion

3.0 Monetary Theory of Exchange Rate Determination

3.1 Theories of Exchange Rate Determination 3.2 Monetary Models of Exchange Rate Detemination

Flexible Price Monetary Approach Sticky Price Monetary Approach

3 3 The Sticw Price Mode1 and Ghana's Long-Run Exchange Rate Uncovered Interest Rate Parity Purchashg Power Parity Money Demand Fmction

3.4 Conclusions

4.0 Data and Mode1 Estimation

4.1 Data Description 4.2 Order of Integration of Ghaua-US Series 4.3 Engle-Granger Two-Stage Procedure 4.4 Johamen Procedure

5.0 Conclusion

Bibliography

List of Figures

3.1 Sequentiai Chow Test for Money Dernand Function

3.2 Forward Recursion CUSUMSQ for Money Demand Fmction

4.1 Logarithmic Dineremes of Levels of Macroeconornic Senes

4.2 Sequentiai Chow Test for ECM

4.3 Forward Recursion CUSUMSQ for ECM

List of Tables

Monetary Indicaton

Exchange Rate and Real GDP Growth

Balance of Payment, Exports and Imports

Uncovered Interest and Real Interest Dif5erential:Ghana-US

Augmented Dickey-Fuller Unit Root Test Results

PhiUips-Perron Unit Root Test Resuits

Engle-Granger Test for Cointegration

Asymptotic Critical Values for Cointegration Tests

Error Correction Mode1 Estimates

Johansen's Likeiihood Ratio Test

Abstract

The Iong-nui relationships between the Ghana-US bilateral exchange rate and relative money

supplies, real incornes, and interest rates are exmineci for the penod 1983-1 994 based on the

sticky pnce monetary model of exchange rate determination. The Engle and Granger, and

Johansen cointegration methodologies are employed. As a coroliary of the former

methodology, the cointegrated system is reduced to a simple error correction model (ECM).

The ECM is subjected to a battery of diagnostic tests such as stability, parameter constancy,

heteroskedasticity, and autocorrelation, and pefioms reasonably well. The ECM M e r

shows that money supply and interest rates are unimportant in the short-run, but the Bank

of Ghana should account for them when fomulating Long-nui exchange rate policies. The

finding of one cointegrating vector is evidence that the model is a representation of long-nui

equilibnum between the exchange rate and the monetary variables in Ghana.

Acknowledgement

1 am grateful to Professors Kuan Xu, Taian Iscan and U.L.Gouranga Rao for helpful comrnents and suggestions. 1 wodd also like to thank Mr. Dennis Canterbury for his insights and advice. AU errors are entirely my responsibility.

Chapter 1

Introduction

The Ghanaian economy experienced severe crisis during the 1970s and early 1980s. The

main contributory factors for the crisis emanated fiom: a fixed and highiy over-valued

exchange rate; large deficits in the Governments budget; and inappropriate monetary

expansion which resulted into inflationary pressures, and imposition of price controls over

a wide range of goods. A prominent feature of the economic crisis was the inappropriate

policy measures with respect to the exchange rate. The government of Ghana pursueci a k e d

exchange rate policy in the 1 WOs, which in part caused the over-valuation of the Ghanaian

currency, and thereby a dramatic decline in exports. The fa11 in exports which resulted in a

decline in foreign exchange earnings had significant eEects on the entire Ghanaian economy.

This policy was also partly responsible for lower levels of production and productivity,

increased poverty, and excessive govemment intervention in the economy.

In 1983 the Ghanaian government was left with Little choice but to hplement a

structural adjutment and economic recovery programme, under the guidance of the

IMF/World Bank formula chiefly in the areas of exchange rate adjustments, and reduction

of public sector deficits. The exchange rate policy of the adjustment and recovery

programmes dowed for the value of the Ghanaian currency to be detemiined by market

forces. But ever since the floating exchange rate policy was implemented in 1983, as s h o w

2

by the economic statistics, output, and balance of payments have experienced wide

fluctuations, and inflation has been consistently high.

The problem is that there was an apparent absence of a valid fbrnework for analyzing

exchange rate movements in Ghana This problem was refiected and reinforced by the

Ghanaian monetary authorities use of what could be termed an ad hoc approach to

formulating exchange rate policy. An ad hoc exchange rate approach is denned as one which

draws on several aspects of exchange rate theories without an explicit reliance on any

pxticular theoreticd mode1 appropriate to Ghana's economic conditions. This means that

exchange rate policy in Ghana is formulated on an eciectic basis by drawing on different

aspects of the variety of exchange rate theories identifïed in the literature. For example, in

order to improve Ghana's extemal accomt, the government devalued the Cedi in 1983. This

was essentially a balance of payment (BOP) measure. The use of this approach was

discontinued by 1986 because the initial over-vaiuation of the Ghanaian currency was

rectifed. In the p s t 4 986 period, the government employed a somewhat monetax-y approach

through the use of interest rates and innuencing the money supply. The objective was

essentidy to stabilize the value of the currency. But it is possible for both devaluation and

interest rate increases to have opposite effects on the value of the currency, i.e., while

devaluation increases the home currency price of foreign currency, interest rate increases

may cause it to fd. Accordingly, it is imperative that the authorities select a fhmework most

appropriate to Ghana's economic environment because this will enable them to better plan

for the future, and also to maintain some stability in the value of the Ghanaian currency.

Clearly, there is no single exchange rate mode1 which wili provide a panacea for al1

3

the problerns of long-run exchange rate determination in Ghana It can be argued that in

mderdeveloped economies like Ghana, the monetary authorities need to select a model of

Iong-nin exchange rate determination which is suitable to their economic conditioos. The

sticky price monetary model is a likely candidate. The ad hoc approach of the Bank of Ghana

was not very helpful for solving or alleviahg the country's pressing problems of exchange

rate management because the Bank's approach did not allow policymakers to search for a

model suitable to Ghana's economic environment. It wiii be shown below that such a

situation had devastating consequences for the Ghanaian economy. For instance, after more

than a decade of the ad hoc approach, the exchange rate, and pnces have both increased by

more than 1000%.

The objective of this thesis is to axertain whether the sticky price monetary model

can uncover any meaningful long-nui relationships between the Ghana-US bilateral

exchange rate, and relative money supplies, interest rates, and real incornes. Specincally, the

thesis examines the sticlq price monetary model for the Ghana-United States exchange rate

determination, c o v e ~ g the period 1983-1994. This model is proposed as one possible

approach that the monetary authorities in Ghana can use to formulate long-run exchange rate

policy. The thesis proposes that the use of this model will not ody address the problem of

"ad hocegt1 in inhanaïan exchange rate management, but will also provide policymakers with

a more appropriate fkmework for policymakhg purposes.

The thesis suggests that the sticky pnce monetary model offers a better explanalion

of the long-m relationships between the exchange rate, and the monetary variables, than

does the current ad hoc approach employed by the Bank of Ghana The sticky price monetary

4

mode1 for Ghana is proposed and appiied Understanding the overd behaviour of the

monetary variables is crucial to long-run exchange rate management. This wili ultimately

lead to the formulation of policies that wiii stimulate economic growth. This study is

conducted within a monetary fiamework An empirical methodology is employed based on

data obtained through archivai and Library searches, and officiai poiicy documents supplied

by the Government of Ghana

The organktion of the remainder of this thesis is as follows. To underscore the

economic decline associated with the ad hoc approach used by the Ghanaian Central Bank,

a description of the Ghanaian economy is provided in chapter 2. In chapter 3, the relevant

literature on the monetary models of exchange rate determination is reviewed, and the

relative strengths and wealmesses of the monetary models are examined within the context

ofthe Ghanaian econorny. The statistical properties of the data, the estimation of the sticlq

price monetary mode1 for Ghana, and the empirical evidence are presented in chapter 4. In

the £inal chapter, the conclusions are summeed.

Chapter 2

Adjustment Programmes and the Ghanaian Economy

Since independence in 1957, and particdarly during the 1960s, the economy of Ghana

witnessed relative prosperity mainly due to high export prices for cocoa, gold, and tirnber,

and a fairly stable political and macroeconomic environment mstry of Finance &

Ewnomic Planning (MFEP) 199 11. Real GDP increased by an average of 3.4% per annum

in the period of 1957-73 (ibid.1991). With a population growth of 2.2% per annum in the

same penod, the annual average growt. of real GDP per capita was 1.2% (ibid.1991).

Merchandise exports increased by 3.5%, while imports decreased by 3 -3 % (ibid. 1 99 1). For

the same period, the Consumer Price Index (CPI) inflation rate averaged 6.3% per annum

(ibid.1991). But fiom 1974 to 1982 however, the Ghanaian economy witnessed an

unprecedented decline. This decline was ascribed to an interplay of several wrongful

policies, namely a fked exchange rate policy which resulted in a highly over-valued

exchange rate that discouraged exports, expansionary fiscal policies that led to large

budgetary deficits, and expansionary monetary policies which resulted in idationary

pressures.

During the decade immediately preceding 1983 when economic adjustrnent was

introduced, the Ghaoaian economy was characterized inter alia, by budgetary deficits, a high

rate of monetary expansion and inflation, declining GDP and export levels, an over-valued

6

exchange rate, and an isolated domestic money and capital market (Kapur 199 1). Ail of these

characteristics persisted dirring the eni of economic adjustment, with the exception of an

overvalued exchange rate (ibid1991). The government financed its huge deficit principally

through h ~ ~ w i n g h m the local financial houses, especiaily the commercial banks, and by

printing more Cedis (ibid. 1 99 1).

The ensuing overview suggests that Ghana's economic decay happened in a penod

in which there was no explicit exchange rate policy. The absence of such a poiicy is partly

responsible for the economic deche in Ghana. Expansionary fiscal and rnonetary policies,

and iIl-planned agricultural, and industrial policies were other explanatory factors.

2.1 Economic Adjustment Programmes

The structural adjustment and economic recovery programmes in Ghana have their

theoretical origins in the classical idea of the k e market as the most efficient ailocator of

scarce resources. In essence, the policy prescriptions of these programmes are based on the

neo-classical theory of supply and demand management. The main goal of demand

management is to r e h c t the level of effective demand by suppressing the wage rate,

lowering the money supply, and restricting credit by adjusting the interest rate (ibid. 199 1).

Supply management is designed to stimulate output by lowering production costs

(ibid.1991). The following discussion enurnerates the salient features of the adjustment

programme and the exchange rate policies contained therein.

Economic Recovery Programme

Ghana's economic recovery programme has been undeltaken in a number of phases. These

phases are: Phase 1,1983-84; Phase II, 1987-88; Phase III, 1989-90; Phase IV, 199 1-3; and

Phase V, fiom 1994 to present. Ghana's Economic Recovery Programme (ERP) was

conceived of as an on-going process to place:

The Ghanaian economy on a sound recovery fiom three essential structural bottlenecks/rigidities, namely (1) colonial economic structure, (2) monocultural export structure and (3) weak moral, social, and physical hfkstmcture (MFEP 1991).

The broad objectives of the ERP were to:

(1) shift relative pnces in favour of production, particdarly for exports; (2) restore fiscal and monetary discipline; (3) initiate rehabilitation of the country's productive base and its economic and

social infrsistnicture; and (4) restore incentives for pnvate savings and investment (ibid. 1991).

According to MFEP (1991), the following specific objectives were forrnulated:

1. To reform the exchange rate system and render it more flexible; . . il. To refom the customs t a f i system in order to simpli& the rates to a more

... d o r m structure; m. To establish realistic relative prices and incomes in the context of the large

achievements in the exchange rate; iv. To restore fiscal discipline; v. To phase out the reduction of the extemal payment arrears which stood at

USS6Ol million at the end of 1983, in order to re-establish the country's credit worthuiess; and

vi. To formulate programmes and projects with a view to rehabilitating the key sectors of the economy including cocoa, gold, timber and mining as part of the 1984-86 ERP. These programmes included sector specific measures designed to restore incentives, improve management, ensure adequate inputs and replace capital.

Since the ERP was implemented in 1983, a number of goals were achieved. F i a the

depreciation of the exchange rate resulted in a doubling of the producer pnce of cocoa

8

Second, administered pices fidly adjusted to the devaluation. Finally, there was a decline

in inflation and positive real interest rates were attained in 1984 (ibid. 1 99 1).

Exchange Rate Policies

As was mentioned above a key gain of the ERP is the liberalkation of the exchange rate. The

specinc measures employed to achieve that goal were: a devaluation of the Cedi; the

introduction of a foreign exchange auction at the Centrai Bank in which individual exporters

made bids; the unification of the black market and official exchange rates by the introduction

of Foreign Exchange Bureaus which were allowed to operate at market prices; and the

unification of the auction and the inter-bank markets by allowing commercial and

development banks to hold auctions as weU.

2.2 Financial Structure

Ghana's financial system comprises a number of banking and non-bank financial

intermediaries. The Bank of Ghana (BOG) is the monetary authority, which in conjunction

with the Mini- of Finance and Economic Planning regulates the behaviou. of the entire

banking and non-bank £inancial sectors. MFEP has the broad fiuiction of fïnancially

administering the country. This inter alia, entails the prudentid management of the Ghana's

finances with a view to ensuring its soundness. As a consequence, it formulates fiscal and

monetary poiicies in the discharge of its duties. In addition, it is also the oniy agency

authorized to contract loans on behaifof the Government of Ghana The Bank of Ghana is

9

empowered by the Bank of Ghana Act to implement fiscal and monetary policies formulated

by MFEP. It perfonns several fiinctions in the execution of its duties. It is responsible for

printing Ghanaian currency (notes and coins). The Bank also has regdatory oversight over

the entire banking and non-banking sectors. It also s u p e ~ s e s banks, by seaing uniform

accounting and auditing standards. The Bank also manages Ghana's foreign exchange

reserves, and in addition is accountable for maintahhg stability in the value of the Ghanaian

currency. In this regard, it buys and sells foreign currency. The banking and non-banking

sectors are also regulated by statutes, namely the Bankùig Law 225 and the Securities

Industry Law. The Banking Law provides a sound and prudentid legal framework for

Ghana's banking sector. It requires banks to maintain a minimum risk-adjusted capital base

equivalent to 6% of net assets. In addition, it sets equable accounting standards and places

restrictions on rkk exposure to single sectors and customers. The Secunties Ind~stry Law

stipulates the establishment of a S e c d e s Commission to regulate the activities of players

in the securities indusûy. These include the stock exchange, investment dealers, and stock

brokea. The Law m e r empowers the Commission to ensure that dealings in securities are

fair, orderly and transparent.

The Central Bank employs interest rates and money supply as instruments of

rnonetary policy. Historically, the Bank of Ghana centred its attention on influencing the

money supply growth and directed monetary policy to that purpose. Its primary tool was its

open market operatiom. If the Bank thought the money supply was too low, it would

purchase bonds and treasury bills util the money supply rose up to the target range. If it

thought the money supply was to hi&, it would sel1 bonds and treasury bills to the general

10

public and fimuciai institutions. The use of required reserves as an instrument of control is

less popular. This is simply because it is solely directed towards h c i a l institutions

(maidy banks), and as such excludes the general public. The Bank therefore sees open

market operations as an effective tool for controlling money supply growth. The Bank of

Ghana since 1983 mostiy focused monetw poiicy on quantitative credit controls. This took

the form of credit ceilings to both the public and private sectors. This policy was gradually

abandoned because the government had initiated a process of h c i a l deregdation in 1986.

Consequently, a market-oriented system of monetary control was initiated by 1987 tbrough

the use of market-detemiined interest rates. In 1989 when inflation had reached very high

levels by Ghanaian standards, the Bank of Ghana changed it policy and switched its target

fiom interest rates to the rate of monetary growth. As a result, monetary policy concentrated

primarily on targeting the money supply and influencing its transmissions throughout the

economy (Kapur 199 1). Nevertheless, these policies did not prove f U y supportive of the

exchange rate poiicy because the rate of rnonetary expansion still remained high and real

interest rates failed to reach satisfactory levels (ibid. 1991). Monetary policy in the post-1989

era has continued to focus on controliing money supply growth principally through open

market operations.

There are three large commercial banks in operation in Ghana - the Ghana

Commercial Bank, the Standard Chartered Bank of Ghana, and the Barclays Bank of Ghana.

The Ghana Commercial Bank is a public enterprise, while the latter two banks are branches

of multi-nationals. The commercial banks in addition to taking deposits and making loans,

undertake money market and foreign exchange activities (buying governrnent securities,

11

issuing certificates of deposits, and buying and s e h g foreign exchange). Further, there are

seven secondary or deveiopment banks in operation in Ghana. These banks are the Social

Security Bank, the Bank for Housing and Construction, the National SaWigs and Credit

Bank, the Agricul- Development Bank, the National Investment Bank, the Merchant

Bank, and the Bank for Credit and Commerce. AU of these banks, with the exception of the

Bank for Credit and Commerce, are public enterprises. These development banks are mainly

involved in commercial banking, although they are supposed to perform a developmental

function, i.e., making project loans to the various sectors. In addition, they also engage in

money market activities (mainly by investing in govemment securities), and buy and seii

foreign exchange. These institutions are required by regulation to hold a certain amount of

required reserves as determined by BOG. The Bank of Ghana, for example may increase the

reserve requirement and aIlow the commercial banks to hold part of their reserves in treasirry

secuities. These banks by complying with this directive play a crucial role in making

monetary policy successful. Also by v h e of the fact that they engage in money market

activities means that the effectiveness of the open market operations depends to a large

extent on their participation. There are three merchant banks in the financial system. These

banks are Ecobank Ghana, Continental Acceptances, and Meridian Bank. These banks are

joint-ventures between the govemment of Ghana, and local and foreign shareholders. They

concentrate on corporate finance, advisory services, and money and capital markets activities

(underwriting, stock broking and investing in govenunent securities). Featuring among the

merchant banks as well is a small CO-operative bank, and over 100 small rural banks

established to mobilize resources in the rurai areas (Kapur 1991). Like commercial and

12

development banks? these merchant banks are required to hold a certain amount of reserves.

They also invest in government securities. Thus, it is an undeniable fact that they play a very

signincant role in ensuring the success of monetary policy.

The non-bank financial sector comprises inter dia, two discount houses, Le., the

Consolidated Discount House, and Securities Discount Company, the Social Security and

National Insurance Trust (SSNIT), the Ghana Stock Exchange, the Home Finance Company

WC), the Export Finance Corporation (EFC), and about 20 insurance companies, including

the State Insurance Corporation, which is the largest. The discount houses were established

in 1988 and 1991 to enable banks better manage their liquidity position. They mainly accept

short-term deposits fiorn banks and other financial institutions, and also invest in shoa and

medium term govemment securities. They are required to hold a large chunk of their assets

(about 70%) in short-term securities. The SSNIT is a public institution mandated to collect

social sec- contributions and pay social s e c m contributions to participants upon

retirement. It is also dowed to invest up to 6% of its assets in govemment securities. The

HFC and EFC specialize in mortgage b c i n g and export b c i n g respectively. These

non-bank financiai institutions by engaging in the purchase of govemment securities, also

play an important role in the centrai bank's open market operations directed at them. The

discount houes and the banks are key players in the foreign exchange market. Less

important players are the insurance companies. It is worth remarking that the npid

depreciation in the value of the Ghanaian currency in the post-1983 era affected the entire

financial sector, mostly banks, adversely. For example, the external iiabilities of most banks

escalated excessively making them fïmmcially worse off.

2.3 Economic Review (1983 -1994)

The ensuing economic review discusses monetary indicators, exchange rates, gross domestic

product, and balance of payments, imports and exports.

Monetary Indicritors

A w e y of the monetary indicators shows that growth in the money supply is considerably

high despite the fact that it has sloweddown since 1 983 (Table 2.1). The high rate of growth

in the money supply was due to excessive govemment borrowing in the domestic capital

market to hance its non-productive activities, such as the payment of wages to civil

servants, which was about 65% of govemment revenue in 1992/93. Further, the high rate of

growth in the money supply c m also be athibuted to goveniments printing of excessive

amounts of Cedis to pay its bills (World Bank 1989). In the period of 1983-1994 govemment

borrowing increased by 1 100%.

The slow-down in the rate of growth in the money supply fiorn 1989-1990 was due

m a d y to contractionary monetary policies. The average rate of inflation fiom 1983-1988

was considerably high (Table 2.1). This was due maialy to excess demand created by the

increase in the money supply which went to pay public sector wages (ibid. 1989). However,

the rate of inflation declined sharply fiom 37.1% in 1989 to 18% in 1990. This rate fell

M e r to 10% in 199 1. The welcomed fdl in the rate of inflation was short-lived, because

agricultural production declined and money growth rose. From 1992 to 1994 the average rate

Table 2.1 Monetary Incikators

Inflation Money (Ml)

1983 40.0% 60.5%

Year

Source: Computed from International Financial Statistics Yearbook, 1995 gines 64,34 and 60c)

of inflation soared upwards to 24.8%. The initial fd in the da t ion rate could be attributed

to an increase in food production associated with the agriculturai policies of the economic

adjustment prognimme (MFEP 199 1, and Berry 1995). Also, the more recent inflationary

CPI1

Of the itemized composition of the Consumer Price Index, local foodstuffs account for approximately 49%. The reminder is accoimted for by items such as transportation, dothhg and rent (Kapur 1991).

The interest rate is the rate on the 9 May Government of Ghana Treasury bills.

Growth of Interest

15

trend was due to a decline in agricuihiral output (ibid- 1999, and to the rapid depreciation of

the exchange rate resulting fiom the hi& rate of monetary expansion (Economist Intelligence

Unit (EIU) 1995). Since imports account for a signiscant proportion of the inputs used in

domestic agriculture, the depreciated exchange rate waich resulted fkorn the policy to "fioat"

the Cedi, increased the production costs of small fhmers. The result was a fd in agricuitural

production.

Nominal interest rates were volatile during the period under consideration (Table

2.1). The interest rates increased respectively, fiom 13% in 1983 to 2 1.7% in 1987. However,

in 1988 and 1989 the interest rate fell slightly to about 20.0%¶ but resumed its upward trend

in 199 1 and 1992. It peaked in 1993 at about 3 1 % compared with 13 .O% in 1983. The ex-

post real interest rate was negative for most of the period.

Exchange Rate

The data reveal a dramatic depreciation in the exchange rate between 1983 and 1994 (Table

2.2). Following the introduction of the floating exchange rate in 1983, there was a 990%

devaluation in the Cedis-US$ exchange rate from its h e d value of 2.75 Cedis=lUS$ in 1978

(World Bank 1989). In 1983,3O.O Cedis was equal to one US dollar. By 1987 the rate was

176 Cedis=IUS$, and by the end of 1994 it was 1300 Cedis=lUS% (Table 2.2). The

instability in the value of the Ghanaia. currency has been attnbuted to the relatively high

monetary growth and innationary pressures which prevailed during the period (EIU 1996).

Gross Domestic Product

The components and breakdown of GDP are as foilows: Agriculture, Forestry and Fishing,

47.6%; Mining and IndUStfy, 16%; Manufacturing, 8.5%; and Services, 36.4% (MFEP 1991

and World Bank 1995). Apart from cocoa, the major agricdtural crops produced in Ghana

fall into either of two categories, namely cereals, and starchy staples. The former includes

maize, nce, millet, and guinea corn. The latter comprises food items such as cassava, yam,

Table 2.2 Exchange Rate and Real GDP Growth

Exchange Rate (CedisRIS$)

Real GDP Growth

Source: Computed fkom International Financial Statistics Yearboo k, 1 9 9 5 e e s ae and 99b.p)

17

cocoyam, and plantain- The services sector consists principdy of hotel, restaurant, and

banking operatiom. Manufàcturing can be classified into textiles, wood processing, vehicle

assembly, and food processing. The mining and industry sector consists mainly of mining

activities. Ghana's economy grew at an appreciable average annual rate of 4.55% during the

period under review (Table 2.2). The highest growth rate was recorded in 1987 (5.6%), while

the lowest occurred in 1983 (2.6%). The positive growth rate throughout the entire period

was attributable to increases in the agricultural and seMces sectors (ibid.199 1 and EIU

1996).

Balance of Payment, Imports and Exports

Ghana's irnports consists mainly of capital goods, intemediate goods, fuel, and consumer

goods (ibid. 1996). Exports on the other hand, consists chiefîy of gold, cocoa, and timber

(ibid.1996). Its major principal trading partuers are United States, United Kingdom,

Germany, and Togo. Together they account for approximately 54% of all exports

(ibid.1996). The remaining 46% is accounted for by about 46 countries, such as Japan,

SwitzerIand, Netherlands, and France (United Nations 1994). On the imports side, Nigeria,

United Kingdom, Italy, Japan, and the United States account for approximately 60% of ail

imports. The remaining 40% originates from countries such as Belgium, France, Canada,

Korea, and China, and some 40 other countries, the principal ones being Ivory Coast, and

Togo (ibid. 1 994).

The data on the value of imports and exports show that in general the former grew

at a much faster rate than the latter during the period under review (Table 2.3). Imports

Table 23 Balance of Payment, Esports and Imports (millions of US Dollars)

Source: Computed fkom International Financial Statistics Yearbook, 1995 (Iines 78cbd, 78aad, 78abd, and 74r)

Year

1983

1984 "

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

exceeded exports in every year with the exception of 1984 and 1986. But, although expoa

eamings were greater than the value of imports in 1983 the overall balance of payment was

negative. However, in 1985, and between 1987 and 199 1 although imports exceeded exports

the overall BOP was positive. Further, in 1983, and between 1992 and 1994 the value of

imports was greater than export receipts, and the overall BOP was negative (Table 2.3). The

BOP surplus between 1984-1 985, and 1987-1991 was attributable mainly to the increase in

Balance of Payments

-1 80.9

35.6

14.1

-60.8

140.1

181.1

155.6

f 05.3

109.2

-56.2

-232.5

142.6

Exports

439.1

565.9

632.4

773 -4

826.8

88 1 .O

807.2

890.6

997.6

986.4

1064.0

1227.0

Export Prices (Cocoa)

8.1%

18.7

24.3%

43 .O%

8 1.7%

93.9%

92.0%

100.0%

1 15.7%

1 17.3%

140.0%

248.0%

hports

499.7

533.0

668.7

7 12.5

951.5

993.4

1002.2

1 198.9

13 18.7

1456.7

1728.0

1580.0

19

exports (Table 2.3), and improved export pnces (Table 2.3). The flexible exchange rate

strategy p m e d by the Govemment since 1983 was aiso partiy an explanation for this trend

(Kapur 1991).

2.4 Conclusion

Several observations may be made from this brief description of Ghana's financial system,

and review of its economy. The feature of the hancial system is that it is dominated

by public enterprises. It appears that the state sector has "crowded-out" the pnvate sector in

the financial system. Also, the financial system appears to have suffered f?om "weak bank

supervision and regdatory ~ e w o r k " (Kapur 199 1). The Government has agreed under the

Financial Sector Reforms to implement more appropriate regdatory masures of the sector.

Another worthwhile observation is that the depreciation of the Ghanaian currency affected

the financial sector (especially banks) davourably.

The £kst two features, namely a predominant public sector, and a weak bank

s u p e ~ s i o n and regulatory fiamework contributed significantiy to the demise of the banking

sector in the pst-1 983 penod. The predicament of the sector was M e r aggravated by the

rapid depreciation of the Ghanaian currency. To be precise, the hanciai heaith of domestic

banks detenorated either because, the exchange rate adjustments soared their extemal

liabilities or rendered some of their customers (mainly public enterpnses) financiaüy

distressed (ibid. 199 1). In addition, the inadequate bank supe~s ion and regulatory regime

worsened the problems of the banking sector, because of the absence inter dia of, W o r m

20

and appropnate accounting standards (ibid. 199 1).

The economic review also reveals severai important features of the Ghanaian

economy. F w the monetary indicators reveal that money growth, and inflation continue to

be relatively high and variable. As a consequence, exchange rate depreciation is rapid, p d y

accomting for the misfortunes of the banking sector. Second, although imports exceed

exports in a number of years, the BOP remained positive. Third, the economy recorded a

positive cyclical growth pattern through out the entire period under study. Finally, interest

rates have been volatile for majority of the period under review. The resdts of the economic

sunrey suggest that there is a need for a theoretical h e w o r k to assist policymakers to

account for the behaviour of the exchange rate in Ghana.

Chapter 3

Monetary Theory of Exchange Rate Determination

In this chapter the literahue on the monetary models of exchange rate determination is

reviewed. The discussion is organized into four sections. Fht , the theories of exchange rate

determination are examined. Second, the monetary model of exchange rate d e t e k a t i o n is

analyzed. Particular emphasis is placed on the flexible, and sticky price monetary

approaches. Third, the applicability of the sticky price monetary model to the Ghanaian

economy is discussed with reference to the uncovered interest rate parïty condition,

pinchashg power parity condition, and the money demand fhction. It is concluded that the

sticky price monetary model is appropnate for the evaluation of the long-mu exchange rate

in Ghana.

3.1 Theories of Exchange Rate Determination

The theories about the exchange rate detemination foliows two distinct traditions: namely

the balance ofpayments (BOP) approach; and the asset approach (McDonald and Hallwood

1994). The asset approach advocates the monetary, and portfolio balance models. The two

major versions of the monetary approach are the flexible, and sticky price rnodels.

The BOP approach to the exchange rate detennination focuses on some major factors

of the balance of payments. According to the BOP approach, the exchange rate is

determined by the flow of currency through the foreign exchange market (Solnik 1996). This

view is Iimited because its main focus is on the trade account due to the historid restriction

on capital flows at the t h e the theory was fomdated. Capital flows are treated as

exogenous shocks rather than being endogenous to the mode1 (McDondd and Hallwood

1994). According to the balance of payments approach, a trade deficit wotdd lead to a

reduction in the country's foreign reserves, and lead dtimately to a depreciation of the home

currency (Solnik 1996). In turn this depreciation would improve the terms of trade

(ibid. 1996).

The BOP approach contrasts sharply with the asset view of the exchange rate

determination (ibid. W4). As was mentioned above, this latter approach is sub-divided into

the monetary, and portfolio balance approaches. In both cases the exchange rate is treated

as the pnce of an asset, Le., the relative price of two monies (ibid. 1994). Mussa (1 982) for

instance argues that:

In analyzing the determination of the exchange rate, one should use the tools which are normally used for the determination of other asset prices, such as bonds and share prices.

In both flexible, and sticky price monetary models, non-money assets are assumed to be

pertèct substitutes, and the exchange rate is determined by relative excess money supplies.

The portfolio balance models advocate that non-money assets are imperfect substitutes, and

play a crucial role in the determination of the exchange rate.

3.2 The Monetary Models of Exchange Rate Determination

Before a detailed examination of the flexible, and sticky price models is undertaken in this

23

section, a brief description of monetary approach is presented Exchange rates have been

charactexized by extreme volatility ever since the advent of the "float" in 1973 (Mosa 1994).

This has engendered a great deal of interest and research in exchauge rate detemination. The

flexible, and sticky price monetary approaches, have emerged as important tools for

exchange rate analysis. They derive their name fiom their assumptions which are associated

with the monetary approach to balance of payments (Johnson 1973). Their assumptions are

that there exists a stable money demand function, uncovered interest parity condition, and

purchasing power parity condition; and that the fluctuations of the exchange rates, and

interest rates fluctuations are the main sources of monetary disturbances (Baxter 1994).

However, they m e r on the mechanisms by which they influence the exchange rates, and

interest rates. Monetary theorists dso suggest that in the absence of substantial transaction

costs, the law of one pnce, i.e., purchasing power parity and interest rate parity, will hold in

international markets @ilson 1978). In addition, ail versions of the monetary models assume

that there are no transaction costs in capital markets, and that there are no obstacles to capital

mobility.

Although, the monetary are used extensively in existing empirical work (McDonald

and Taylor 1994), the empincal r ed t s are mixed (Sarantis 1994). Early tests of these models

by Bilson (1978), Hodrick (1978), and Frankel(1979) were supportive of the flexible price

monetary model. A number of research examined the monetary models for the post-1978

period, and indicate that the models did not perform weli (McDonald and Taylor, 1992).

Perhaps the most serious challenge to the monetary models is that they are not used to

represent a long-run equilibrium between the exchange rate and the monetary variables

(Meese 1986; Boothe and

explanation for these dismal

24

Glassman 1987; and McNown and Wailace 1989). One

results relates to the fact that the assumptions underlying the

monetary models - stable money demand fiinction, PPP and uncovered interest rate parity -

may not be valid in the real world. Goldstein et ai.(1980) and Aliber (1973) have suggested

that uncovered interest rate parity may not hold in the reality because of various country

risks, lack of market integration, failure of instantaneous PPP, errors in measuring expected

exchange rate depreciation, and rïsk premium. The assumption of a stable money demand

hction albeit appealing, does not lend support to the model in reaiity. The reason is simply

because the underlying economic structures of the two countries may undergo a variety of

changes thereby making the money demand functions unstable over time. Finally, the PPP

concept per se is controversial in the sense that there is not any agreed upon price index to

be used in computing the parity (Frenkel 1976). When the resdts of a theory seem to hinge

cruciaily on an assumption such as the PPP, then if the assumption is dubious, the resuIts are

suspect The debate in the Literature is whether the index shouid pertain to traded goods ody

or whether it should cover the broadest range of commodities.

These criticisms notwithstanding, McDonald and Taylor (1994); McNown and

Wallace (1 994); and Van Der Berg and Jayaneai (1 993) have demonstrated that the rnonetary

models do have some long-nui validity. The study by McNown and Wallace (1994), was

primarily motivated by the works of Abuaf and Jorion (1 !WO), and Choudhry, McNown and

Wailace (1 99 1 ), who found some evidence favourablle to long-run PPP. Of particular interest

is the approach adopted by McDonald and Taylor (1994). They tested for the long-nui

validity of the model using Johansen's cointegration procedure (Johansen 1988 and Johansen

et ai. 1990) and modelled the short-run dynamics via an Error Correction Mode1 (ECM).

33.1 Flexible Price Monetary Approach (FLMA)

The flexible price monetary approach is proposed by Frenkel(1976), Mussa (1976), and

Bilson (1978). It assumes that pnces are fUy flexible, and that purchasing power parity

(PPP) holds instantaneously. It also assumes the existence of a stable monetary equilibrium

between reai money demand and real money supply. Finally the rnoney supply and real

iocome variables are assurned to be exogenously detefinined. The mode1 embodies the

following equations:

m, 'Pt +byt -ai,

m; = p', + by; - aiet,

and

st = Pt - P.,

where m, and met are the logarithms of the domestic and foreign money supplies respectively;

y, and yet are the logarithms of domestic and foreign real incornes; it and i', are the domestic

and foreign interest rates, respectively; s, is the logarithrn of the spot exchange rate (home

price of foreign currency); p, and R* are the logarithms of the domestic and foreign pnce

levels, respectively; and a and b are parameters representing the interest rate and income

elasticities respectively across coutries.

Equations (3.1) and (3.2) are the money market equilibrium conditions for the home

26

and foreign countries respectively, and represent 1ogarit)unic versions of the typical Cagan

(Cagan 19%) demand for money function. Equation (3.3) is the purchashg power parity

hypothesis. Solving for p, and p; in equations (3.1) and (3.2), and substituting into equation

(3 3), and adding a disturbance te- yields the nnal reduced-fonn exchange rate equation:

where g, is the random error tem, and dl other variables are as previously defined. If the

FLMA is correct, it is expected that P,=l, P,4, P, >O.

McNown and Wailace (1 994) argue that:

According to the monetary approach, inmeases in home (foreign) money supply and interest rates, the latter by reducing the demand for money cause a depreciation (appreciation) of the domestic currency. Increases in home (foreign) real income, by increasing the demand for money, cause the domestic currency to appreciate (depreciate) (McNown and Wallace 1994).

The positive effect of changes in the domestic interest rate on the exchange rate, reflects the

fact that an increase in the latter reduces the dernand for rnoney, which in tum leads to a rise

in the domestic price level. In this way money market equilibrium is maintained. However,

given that PPP holds, the domestic price level can only rise if the exchange rate depreciates

(McDonald and Hdwood 1994). A rise in income increases the transactions demand for

money and, with a constant nominal money supply, money market equilibrium can oniy be

restored if the domestic pnce level f d s ; this in tum cm only occur given a strict PPP

assumption, if exchange rate changes (ibid. 1994).

27

3.2 Sticky Price Monetary Approach (SPMA)

This approach is attributed to Dombusch (1976) and Frankel(1979). AU of the assumptions

underlying the FLMA still hold, with the exception of instantaneous PPP. Both Dombusch

and Frankel argue that in the short-nui prices are more iikely to be sticlq, due to "menu

costs" and imperfect information, and as a consequence PPP does not hold instantaneously.

For example, Frankel assumes that the adjustment of the exchange rate to its equilibrium

level depends on the r d interest rate differential, such that PPP holds onIy in the long-run.

Thus, in an environment where PPP does not hold instantaneously, the current exchange will

deviate f?om its long-run Ievel such that the following relationship holds (Sarantis 1994),

where zt and x; are the Mation rates in the home am i foreign corntries respect ively .

Assuming that equilibrium values are given by their current Levels, replacing equation (3.3)

by (3.5) and then substituthg for pt and pt9 fiom equations (3.1) and (3.2), i.e. the monetary

equilibrium conditions and adding a disturbance term, we obtain the reduced-form SPMA

equatio n :

If the SPMA mode1 is correctly specifïed, the coefficients of the variables entering the

FLMA have the same interpretation. However, the fiutdamental distinguishing feature

28

separating the two models is the presence of home and foreign country inflation rates, the

impacts of which are expected to be negative (i.e.P,<O).

3.3 The Sticky Price Model and Ghana's Long-Run Exchange Rate

In this section an anaiysis of the sticlq price monetary mode1 is presented to determine its

suitability for the Ghanaian economy. In order to establish the relationship between the

SPMA and Ghana's economy, an empirical analysis of the uncovered interest, and purchasing

power parities, and the money demand function are undertaken.

3.3.1 Uncovered Interest Rate Parity

The uncovered interest rate parity (UIP) assumes that an investor is indifferent between

investing one doilar in home or foreign treasury bills, or some other financial asset such as

a bank account, without covering the risks of foreign exchange rates changes in the fomard

market, when the expected rates of retum on them are equal and nsk fiee (McDonald and

Haüwood 1994). in other words, the foilowing condition holds:

where gt and ,q* denote the nominal yields to rnaturity on k period domestic and foreign

bonds in period t, respectively; s, denotes the logarithm of the exchange rate, definecl as wits

of foreign currency per unit domestic currency; and E&,, - sJ denotes the expected change

in the log of exchange rate between periods t and t+k (ibid. 1994).

The uncovered interest rate differentiais (UTD) is defhed as the rate on the three

month US treasury bill minus the rate on three month Ghana treasury bill denominated in

Ghanaian c~rrency.~ The UIDs for Ghana-US currencies deviate sipnincantiy fkom zero

(Table 3.1). It may be speculated that the deviation of the UIDs from zero, in the case of

Ghana, reflects a possible level of uncertainty about country risk, i.e. imposition of capital

controls, taxes, or other country specinc restrictions on capital flows, and defauit risk,

especialiy on pnvate securities. This speculation is encourageci by Alibefs (1 973) argument

that political risk is one reason why UIP may not hold.

33.2 Purchasing Power Parity

There are two versions of PPP: the absolute and the relative. The former States that the

equilibrium exchange rate equals the ratio of domestic to foreign prices. The latter relates

changes in the exchange rate to changes in price ratios (Frenkel 1979). McNown and Wallace

(1994) observed that "the h t building block" of the monetary mode1 "assumes long-nin

(relative) PPP". This mode1 assumes that ex ante PPP wodd hold if prices were fully

flexible, i.e.,

See Goldstein, Mathieson, and Lane (1980) for a similar definition of ULD in their discussion of covered and uncovered interest rate differentiais between the US dollar

and the French, German, Japanese, and British currencies.

30

Ifprices are less than M y flexible, as in the SPMA, the= may be temporary deviatiom fiom

the ex ante PPP condition. PPP relations are nevertheless assurned to hold in the long-nin

when price level adjustments are complete. The relative PPP theory predicts that exchange

rates are determined by inflation differentials. This is closely related to the Fisher theory of

interest rates, i.e.

and

where rt is the real interest rate, and Ap;,, is the expected inflation rate over the maturity

horizon of the underlying bond, based on information available at t h e t and asterisks denote

corresponding foreign variables. The Fisher theory predicts that market interest rates change

in response to changes in inflation. The interest rate theory predicts that the disparity

between two courmies' interest rates will reflect the differential between their expected

inflation rates. Therefore the exchange rate of the high-inflation country would depreciate

enough to wipe out the advantage of its higher interest rate if these parities held exactly

(Spiro 1989).

The real interest rate differentials (RIDs) for Ghana-US are depicted in Table 3.1. The

real interest differential (RD) is defined as the difference between the real rate of interest

31

on 90-day treasury biils denominated in each cunency." The real interest rate in each country

is defined as the 90-day treasury bill adjusted for inflation, as measured by the Consumer

Price Index (Goldstein et al 1980). The data show that there are signincant depamires of RID

for Ghana and the United States between 1983 and 1993 (Table 3.1). It reduced cùasticdy

between 1983 and 1989, but increased sharply thereafter. Deviations of the RID fiom zero

Table 3.1 Uncovered Interest and Real Interest DîfEerentials: Ghana-US

year I IJID I RID

Source: Computed fkom International Financial Statistics Year Book 1995 @ne 60c)

See Goldstein, Mathieson, and Lane (1980) for a simila. definition of RID in their discussion of real interest rate differential between the US dollar and the French, German, Japanese, and British currencies.

can reflect the fdure of either UID or instantaneous PPP (ibid.1980).

3 3 3 Money Demand Fonction

The monetary models assume a stable money demand fiinction. Foilowing is an empiricai

test for the stability of the money demand h c t i o n in Ghana The mode1 which is tested

closely replicates equation (3.1):

where Bo, Pi and p2 are parameters and a i l other variables are as previously defined.

Two tests (Chow 1960, and Brown et al. 1975) are employed to determine the

stability of Ghana's money demand function. The former provides a means for determining

parameter constancy, whereas the latter is used in detecting a structurai break The Chow test

divides the sample into two parts. Part one is the baseline period, while the second comprises

the observations from the end point of the baseline period to the last observation used in the

recursive estimate (White 1993). The Chow Test is caiculated using the following formula:

where SSEl and SSE2 are the sum of squared errors fiom the first and second parts of the

split sample, N is the number of observations in the entire sample, K is the number of

33

estimatecl parameters, N, and N, are the observations in the fkst and second part of the split

sample respectively, and SSE is the residual sum of squares fkom the regression over the

entire sample. Brown et d(1975) suggested that in order to ascertain the extent of a

structurai break within a model, the cumulative sum of squared residuds defined as

needs to be computed, where vj and v, are the residuds fiom the regression and k is the

number of parameters. This test is widely known in the econometric Literature as

CUSUMSQ. The results of the Chow statistic is shown in Figure 3.1. Here the confidence

interval at 95% is provided in order to detect any depamires fiom constancy. There is strong

evidence of a break fiom parameter constancy in 1983 and 1988. Figure 3.2 shows the

recursive CUSUMSQ at the 5% level of signincance. Any deviation from the upper and

lower bounds is evidence of instability in the model. Note that there is a break fiom stability

over the entire sample. This instability can be associated with those periods during which

prices, money supply, GDP and interest rates experienced wide fluctuations (see Tables 2.1

and 2.2). Both the sequential Chow and CUSUMSQ tests reveal that the money demand

fiuiction is not stable across the penod under study (1983-1994). Put simply, these tests

suggest that the parameters of the model are not invariant over tirne.

34

3.4 Conclusion

The conclusion can be drawn fiom the results of the tests conducted in this chapter that the

sticky price monetary mode1 may explain long-run exchange rate behaviour in Ghana. The

empiricd evidence suggests that the money demand h c t i o n is not stable over tirne.

Furthemore, the evidence does indicate signincant departures fiom the parity conditions in

both directions. Clearly, the latter is in conformity with the view that pnces are seemingly

sticky, adjusting graduaily, such that PPP may hold in the long-nui. On this basis, it is

possible that the SPMA will be an approximation of long-run equilibrium towards which

the Ghanaian economy tends to adjust This leads one naturally to wonder whether the

SPMA is a better fit especially given the fact that a number of researchers have expressed

some scepticism about its validity. The question being addressed is whether the SPMA is

more useful in understanding actual exchange rate behaviour in Ghana. As it happens,

however, that question is very controversial, and not one that can be answered in a few

sentences. As mentioned before, the PPP assumption is crucial, Le., one on which the

conclusions depend sensitively. However, fiom the foregoing discussion, and ceteris paribus

the SPMA codd offer a better explmation of the long-nin exchange rate movement in

Ghana.

Chow Statistics

Chapter 4

Data and Mode1 Estimation

The data sources and the estimation redts are reporteci in this chapter. In section 4.1, a bnef

description of the data sources is presented. In section 4.2, testing for unit root are

considered. Sections 4.3 and 4.4 are devoted to the application of the EngIe and Granger two-

stage, and Johansen test procedures for cointegration respectively.

4.1 Data Description

The data of the Ghana-US exchange rate, and Ghana and US macroeconomic variables are

ail taken fiom the International Financial Statistics Year Books (IFSYB) from 1983-1994.

The exchange rate used is d e h e d as (Ghanaian currency per US dollar). The chosen

monetary aggregate is MI, while the real incorne measure is real GDP, and the short-term

nominal interest rate is the three-month treaniry bill rates. The Consumer Price Index is used

as a measure of the inflation rate. All domestic variables, except the short-terni nominal

interest rates and the Consumer Pice Index are measured in Ghanaiao currency. The US

variables, except the short-term nominal interest rates and Consumer Pnce Index are

measured in US dollars. Other sources are fiom the Economist Intelligence Unit (EW)

Country Reports on Ghana (1 992-95), the Ministry of Finance and Economic Planning,

Ghana, and Statisticai SeMce, Ghana.

38

4.2 Order of Integration of Ghana-US Series

Plots of the levels of ail the macroeconomic variables entering equation (3.6) are shown in

Figure 4.1. Apart h m s,, a visual inspection of the levels of &'t-y3, (i:-4). (m*t-rnJ and (xta-

xJ show that these do not display a clear-cut upward trend. In fact it is very difncdt to

visually decipher whether they are nomtationary or not The behaviour of (yvt-yJ, (ivriJ, (mat-

mJ and (z;-q) c m largely be attributed to volatility of interest rates, rapid growth of money

supply, high and variable inflation rates, and output fluctuations in Ghana during the period

under investigation (see chapter 2 for a detailed explanation).

In order to verify statistically if the variables have unit roots, the augmented Dickey-

Fuller (198 l), and Phillip-Perron (1988) tests are performed. Test results appear in Tables

4.1 and 4.2, respectively. In both tables, the results for considering constant trend and

constant no trend cases are shown. The null hypothesis of a unit root in the levels of all the

variables cannot be rejected at the 10% level of signincance? Both the Dickey-Fuller and

Phiilip-Perron tests confïnn the non-stationary nature of the data. Each of the variables

requires only one diEerencing in order to remove the nonstationarïty. This is evidence that

all the variables are integrated of order one [i.e.,I(l)].

Additional unit root tests performed on the first clifferences of each times series show that the null hypothesis that the first ciifferences of these series had a unit mot can be rejected. In other words, it is confirmed that the levels of al1 the original t h e series are I(1).

39

4.3 Engle-Granger Two-Stage Procedure

Because the presence of a unit root has been established in the variables that appear in the

model, the presence of cointegrathg vectors is tested. Both Engle and Granger (1 987) tests

and, Johansen and Juselius (1992) are used. The nuiI hypothesis being tested that is: There

is no cointegration in the case of the Engle-Granger two-step method. The Engle-Granger

method for testing for cointegration fïrst involves estimating by OLS, the foliowing

regression

A test for cointegration is given by a test for a unit root in the estimated error term et. This

is achieved by conducting an augmented Dickey Fuller test on the residuals fkom equation

(4.1) The results of this test are presented in Table 4.3. Since in absolute terms the test

statistic exceeds the criticai value, the null hypothesis of no cointegration among the

variables is rejected at the 10% level of significance. The conclusion would be that the

estimated e, is stationary, and hence the variables are cointegrated. Critical values for these

tests appear in Table 4.4. The second stage, which follows fiom the Engle and Granger

representation theorem, irnplies that any set of cointegrating variables which have an Auto-

Regressive Integrated Moving Average (ARMA) representation may be 7Nntten in the form

of an error correction model (ECM). Here, the parameters obtained fiom the regression in

the first stage, are used in an error correction fom. The simple ECM (Engle and Granger

1987) which captures the dynamics of the system can be WRtten as:

where A refers to the fïrst Merence, q-, is the one period lagged value of the r e s idd fiom

the original regression (4. l), the OLS estimate of the e m r te- and v, is the error term. The

original OLS error terni is construed as the equilibrium e m r and is used to link the short run

behaviour of s, to its long-run value. The essence of formulating an ECM is to correct for this

error.

In this regression, A(m',-mJ, A(& -X ), AG ), A($ -p ) capture the short-nui

variations in (met-@, (Y*~-YJ. (iet-it), (x0&, whereas the error correction term q, captures

the adjustment towards the long-run equilibrium. If the estimate of P, is statistically

signiscant, it indicates what percentage of deviation fiom the long-run equilibrium s, in one

period is corrected in the next penod. Table 4.5 reports the esfimated ECM, together with

Lagrange Multiplier test statistics pertaining to the residual serial correlation x2,; and a test

for heteroscedasticity x2, These results indicate that the short-m changes in (y*,-y,) have

a substantial positive impact on s, whereas short-run changes in (m*t-mJ, (iot-iJ, (x;-xJ have

insignificant negative effects on st. Specifically, the short-run effects reveal that a 1%

reduction in the relative money supply and inflation rates leads to a 0.659% and 0.3 16%

appreciation of the exchange rate respectively. On the other hand, a 1% improvement in real

incornes and interest rates results in 1.732% and 0.151% depreciation of the currency

respectively. Note that the coefficient estimates of the interest rate and inflation variables do

41

not have the expected sign(s). In addition, the evidence M e r shows that about 0.72% of

the discrepancy between the actual and the long-run, or equilibrium, value of s, is eliminated

or corrected each qwuter. Alternatively put, disequilibrium in the exchange rate is corrected

through a process in which the achial exchange rate gravitates towards equilibrium at a rate

of 0.72% per quarter. This represents a fairly high speed of adjustment.

The stability of the ECM was evaiuated by the Chow and CUSUMSQ tests. The

sequentiai Chow test at the 5% level of sigrScance for the ECM equation (4.2) is shown in

Figure 4.2. An inspection of the graph reveals that there is strong evidence of a structurai

break between 1984 and 1987. The period in which the break fiom constancy occur roughly

coïncides with an era characterized by volatile interest rates and marked fluctuations in

money supply, real outp* and high inflation rate. The results of the Chow test suggest that

the ECM does not displays stability over the entire data set The CUSUMSQ test (Figure 4.3)

reveals that there was a strikïng deparhue fiom constancy between 1985 and 199 1. The

results of both tests show that the ECM was relatively stable over the period under

investigation. The ECM was next evaluated for heteroscedasticity and autocorrelation. The

Lagrange Multiplier test accepted the nuil hypothesis of uncorrelated residuals. The test for

heteroscedasticity rejects the nul1 hypothesis of homoscedastic errors at the 10% level of

significance. Note that the standard errors are heteroskedasticity-consistent (White 1980).

4.4 The Johansen Procedure

Another procedure employed for testing for the presence of cointegration was that suggested

42

by Johansen (1988) and Johansen and Juselius (1990). This approach is useful for confirming

the Engle and Granger r e d t s of cointegration.

In order to implement this procedure, a lag length must be specined for the Vector

autoregression (VAR) and the orders of integration of the series entering the VAR must be

determined. The Akaike Information Critenon (AIC) is used for choosing the optimal lag

length of three. In Table 4.6, the values of the trace statistics obtained using Johansen's

multivariate maximum likelihood technique for estimating cointegration relationships are

reported. The trace statistic is used to test the nuil hypothesis that there are at most r

significant cointegrating vectors among a system of time series (Le., zero against r

cointegrating vectors). The maximum eigenvalue tests the nul1 hypothesis of r cointegrating

vectors qaiost the alternative of r+1 cointegrating vectors. These test statistics are also

referred to in the econometric literature as iikelihood ratio test statistics. On the basis of these

trace statistics, we may reject the null hypothesis that there are no cointegrating vectors. The

hdings suggest that there is one such relationship. The results also validate the Engle and

Granger finding of cointegration for the system.

Table 4.1 Augmented Dickey-Fulier Unit Root Test Results

Asymptotic Critical Values (Davidson & Mackinnon, 1991)

No Trend Trend Level of Significance

-2.57 -3.13 10%

Variables

S,

Table 4.2 Phiiiips-Perron Unit Root Test Results

Variables

ta(Constant No Trend)

-1 -8 162

Z(t'd(C0flsfant No Trend)

dConstant Trend)

-2.7088

( Z(t*.)(Co~l~fant Trend)

Asymptotic Critical Values (Davidson & Mackinnon)

No Trend -1 1.2

Trend Level of Significance -1 8.2 10%

CRDW represents the Cointegration Regression Durbin Watson

Table 4.3 Engle-Granger Cointegration Test Results

Table 4.4 Asymptotic Critical Values for Cointegration Tests

L

Critical Values

R2

CRDW

l

Test Statistic

dimension = 2 r,(No Trend) r.(Trend)

sa(C0nstant No Trend)

-5.40

-4.13

0.96

1.53

dimension = 3 r,(No Trend) trend)

r,(Çonstant Trend)

-5.70

-4.43

0.96

1.55

dimension = 4 r,(No Trend) trend)

dimension = 5 ra(No Trend) taprend

dimension = 6 se No Trend) ra(Trend

10% Level of Simiificance

Source (Davidson & Mackinnon, 199 1) Note: dimension refers to the number of variables in the mode1

Table 4.5 Error Correction Mode1 Estimates

Variable 1 Coefficients 1 S.E. 1 T-statistics

where R2 is the coefficient of determination, and S.E. refers to the standard errors obtained fiom the regression.

Table 4.6 Johansen's Likelihood Ratio Test

r represents the nurnber of cointegrating vecton. * indicates statistically signifïcant.

a criticai values are at 10% level of signincance and are taken nom Osterwald-Lenum, M. (1 992).

The LR test statistic is computed as follows, n

LR= T Z In(1-4) , where Aj are the canonid correlations between j=r+l

the lagged values of a time series and a change in its value, T is the number of observations, and n is the number of variables in the system.

Chow Statistic

Chapter 5

Conclusions

In this thesis, the Ghanaian economy is examined with respect to fiscal, monetary and

exchange rate poiicies to establish its basic characteristics. An in-depth review of the

theoretical underpinnings of the monetary models is presented in Chapter 3. Additionally,

the tenability of the assumptions of the monetary models for Ghana is evaluated. The

empirical evidence suggests that the use of the sticky price monetary mode1 is more

appropriate. This thesis reports the existence of cointegrating relationships among the

exchange rate, and the monetary variables in the case of Ghana.

Several important results emerged fiom this study. As a preliminary step for testing

for long-run relationships between the exchange rate and the monetary variables, the short-

nin dynamics of the exchange rate is modeiied using an ECM. The results of the ECM reveal

that real incomes impact significantly on the exchange rate a d j m e n t in the short-nui

whereas, relative money supplies, interest rates and inflation rates have less significant

effects on short-nui adjustments in the exchange rate. The coefficient of the error correction

term indicates a fairly hi& speed of adjustment of the actual to the equilibrium exchange

rate. Additionally, the sequential Chow and CUSUMSQ tests show that the ECM is

relatively stable for most of the t h e . The cointegration results provide concrete evidence

that there is one statistically signincant cointegrating vector between the exchange rate and

domestic and foreign money supplies, reai income, short-term interest rate, and the innation

49

50

rates in Ghana The research suggests that the sticky pnce monetary model can explain long-

nin reiationships behveen the exchange rate, and the monetary variables in Ghana The

hding of one cointegrating vector is consistent with the findings of other researchers

notably, McNown and Wallace (1994) who applied the model to Chile, Argentina and Israel,

and Hendrik and Jayanetti (1993) who tested the long-run validity of the model for India and

Sri Lanka.

The findings of this thesis suggest that: (1) the model provides a plausible description

of the long-nm equilibrium exchange rate; (2) the equilibrium exchange rate is not strongly

linked to the underlying instruments of monetary policy (money supply and interest rates)

in the short-run, but is directiy linked to them in the long-m. The sticky price monetary

model may therefore serve as a useful reference in the formulation of long-run exchange rate

policy for Ghana; and (3) non-monetary factors such as speculation may be important in

explainhg short-run exchange rate behaviour.

Dickey, Jansen and Thorton (199 1) have pointed out that the cointegrating vectors

illustrate reduced-form relationships. Consequently, we should exercise caution when

drawing conclusions fiom these relationships because the reduced-form coefficients are not

representative of the underlying structural model. In other words, equation (4.1) faiis to

capture all the action produced by the true underlying structural model.

In temm of friture research, the findings suggest three fhitful avenues. First, attempts

should be made to estimate a dynamic structural monetary model for Ghana to ver* the

nature of the true relationships among the variables. Second, the empirical results show that

51

monetary factors are less important in explaining short-run variations in the exchange rate

of the Ghanaian cunency and that non-monetary factors are more important. The next step

is to investigate this case M e r . Third, the use of alternative definitions of money in

cointegration modelling is worthy of investigation in Ghana.

Bibliography

Aliber, RZ. 1973. "The Interest Rate Parity Theorem: A Reinterpretation," Journal of Political Economv, 8 1, 145 1 - 1459.

Abuaf, N. and P. Jorion. 1990. "Purchasing Power Parity in the Long-Run," Journal of Finance, 45, 1 57- 174.

Baffes, J. and A. Shah. 1994. "CaUSâLity and Comovement Between Taxes and Expenditures: Historical Evidence fkom Argentha, Brazil, and Mexico)' Journal of Development

conomics, 4 4 3 1 1-3 3 1.

Baiiee, R.T. and D. Selover. 1987. "On Error Correction Models: Specification, Interpretation, Estimation," International Joumal of Forecasting, 3 ,434 1.

Bane jee et al. 1993. Cointegraion. Enor-Correction and the Econometric Analvsis of Non- Station- Data: Advanced Texts in Economeûics. Oxford: Odord University Press.

Baxter, M. 1994. "Real Exchange Rates and Real Interest Dinerentials - Have We Missed the Business Cycle Relationship?," Journal of Monetarv Economics, 33,537.

Berry, L.B. 1995. Ghana: A Country St~dv, Washington D.C.: Federal Research Division, Library of Congress.

Bilson, J.F.O. 1978. "The Monetary Approach to the Exchange Rate: Some Empirical Evidence," International Monetary Fund Staff Papers, Vo1.25,48-75.

Bilson, J.F.O. 1978. "The Cunent Experience With Floating Exchange Rates: An Appraisd of the Monetary Approach," Amencan Econornic Associatio~ Vo1.68, No.2,393-397.

Boothe, P. and D. Glassman, 1987. "Off the Mark: Lessons for Exchange Rate ModeLIing," Oxford Econom c Pa~ers, 39,443-457.

Brown, R., J. Durbin and J. Evans. 1975. "Techniques for Testing the Constancy of e Roval Statistical Society Series B, . .

Regression Relationships over Time," Journal of th 37, 149-172.

Bunting, F.H. 1939. "Purchashg Power Parity Theory Reexamined," Soubem Economiç Jo~rnal , 5, N0.3,282-301.

Cagan, P. 1972."The Channels of Monetary Effects on Interest Rates," National Bureau of ic Researcil,"I New York.

Cagan, P. 1956. "The Monetary Dynamics of Hyperinflation," In Studies in the Q e n i e o y of Monev, Milton Friedman (ed.), Chicago: University of Chicago Press.

Cannan, E. 1921. "The Application of the Theoretical Apparatus of Supply and Demand to Units of Currency," E conomic JOUXTE& 3 1, No. 1 24,453-6 1.

Cassel, G. 1928. Post-War Monetam Stabilization. New York: Columbia University Press.

Charemm, W.S. and D.F. Deadman -1992. New Directions in Econometnc Pmtice, United Kingdom: Elgar.

Choudtuy, T., R McNown and M.S. Wallace. 1991. "Purchasing Power P m and the Canadian Float in the 1950s," Review of Economics and Statistics, 73,558-63.

Chow, G. 1960. "Tests for Equality Between Sets of Coefficients in Two Linear Regressions," Econometricg, Vo1.28,59 1605.

Davidson, R and J.G. MacKinnon. 1993. Estimation and Inference in Econometrics, Oxford: Oxford University Press.

Dickey, D.A. and W.A. Fuller. 1979. "Distribution of the Estimators for Autoregressive Time Series With a Unit Roof" Journal of the American StaQstica.1 Assoc - -

iat io~ 84,427-43 1.

Dickey, D.A. and W.A. Fuller. 198 1. "Likelihood Ratio Statistics for Autoregressive Time Series With a Unit Root," Econometrica, 50, 1057-1072.

Dickey, D.A., H.P. Hasza and W.A. Fuller. 1984. "Testing for Unit Roots in Seasonal Time Series," Journal of the Statistical Association, 79,355-367.

Dickey, D.A. and S. Pantula 1987. "Determirhg the Order of DifEerencing in 1 of Business and Economics Statishcs, Autoregressive Processes," Jouma * .

15,455-46 1.

Diebold, F.X. and M. Nerlove. 1988. "Unit Root in Economic Time Series: A Selective Survey," In Fomby, T.B. and Rhodes, G.F.(eds.), Advances in Econometrics: Cointegpition. S~urious Regressions and Unit Roots, Greenwich : JAI Press.

Dolado, J.J., T. Jenkuison and S. Sosvilla-Rivero. 1990. "Cointegration and Unit Root: A Survey," Journal of Economic Survevs.

Dombusch, R 1979. "The Theory of Flexible Exchange Rate Regimes and Macroeconomic Policy," Scandinavian lounial of Economic~, 78, No.2,255-275.

Dombusch, R 1976. Capital Mobility, Flexible Exchange Rates and Macroeconomic Equilibrium In E. Claassen and P.Slain (eds.), Recent Issues in International Monet-

conomics, North-Hoiland.

Dow, J. and I.D. Saville. 1988. A C . . ritiaue of Monetarv Polk : Theory and British

-rience, Oxford: Clarendon Press.

Economist htefigence Unit. 1992-96. Countrv Re~orts on Ghana.

Edison, H. J. and B.D. Paul. 1994. "A Re-assessrnent of the Relationship Between the Real Exchange Rates and the Real Interest Rates 1974-1 990, Journal of Monet- Fx:onomics, 3 1, 1 65- 187.

Engle, R.F. and C. W. J. Granger. 1987. "Cointegration and Error Correction: Representation, Estimation and Testing," Econometrica, 55,25 1-276.

Engle, R.F. and D.F. Hendry. 1989. "Testing Superexogeneity and Invariance," Mimeographed. University of Oxford.

Engle, R.F., C.W.J. Granger and J. Hallman. 1989. "Combining Short and Long Run Forecasts: An Application and Seasonal Cointegration to Monthly Electriciv Sales Forecasting," Journal of Econometrics, 3 9,4562.

Engle, R.F. and B.S. Yoo. 1979. "Forecasting and Testhg in Cointegrated Systems," Journal of Econometrics, 35,143-1 59.

Ford, J.L. and T. Stark. 1967. L O n e and Short Term Interest Rates: An Econometn c Studv, New York: A.M. Kelley Publishers.

Frankel, LA. 1984. Tests of Monetary and Portfolio Balance Models of Exchange Rate Determination, In Exchange Rate Theorv and Practice. John F.O. Bilson and Richard C. Marston (eds.), The University of Chicago Press. Chicago.

Frenkel, J.A. 1979. "A Monetary Approach to the Exchange Rate: Doctrinal Aspects and O . EmpiricaI Evidence," Scandinavian Journal of EconomiçS, 64,610-622.

Friedman, M. 1953. "The Case for Flexible Exchange Rates," In his book, ]Essavs in Positive 0 .

Economics-203. Chicago: University of Chicago Press.

Fry, M. J. 1978. "A Monetary Approach to Afghanistan's Flexible Exchange Rate," ,Journal of Monev. Credit and Banking, Vol. 10,219-225.

Goldstein, M., D.J. Mathieson and T. Lane. 1980. "DeterPUaants and Systematic Consequences of Internationai Capital Flows," In Williams, R.C.(ed-), International Caoia

arkets: Recent Developrnents and Short-Tem Prospects, Occasional Paper-IMF No. 1.

Gonzalo, J. 1994. "Five Alternative Methods of Estimating Long-Run Equilibrium Relationships," Journal of Econometrics, 60,203-23 3.

Hall, S.G. 1986. "An Application of the Engle-Granger Two Step Estimation Procedure to United Kingdom Aggregate Wage Data," Oxford Bulletin of Economics and Statistics, 48, 229-240.

Hall, S.G. 1989. "Maximum Likelihood Estimation of Cointegration Vectors: An Example . . of the Johansen Procedure," Oxford Bulletin of Economics and Statistics, 5 1,2 13-21 8.

Hall, S.G. 1990. "The Effect of Varying Length VAR Models on the Maximum Likelihood Estimates of Cointegrating Vectors," Scottish Journal of Political Economv.

Hall, S.G., S.G.B. Hendry and J. Wilcox. 1989. "The Long-Run Determination of the U.K. Monetary Aggregates, in Hendry, S.G.B. and Pattenon, K.D. (eds.), Economic Modelling at the Bank of Eneland, London: Chapman and Hall.

Hailwood, C.P. and R McDonaid. 1994. International Monev and F i n a Oxford: Blackweil Publisbers.

Hendry, D.F. and N.R. Erricson. 1993. "An Econometnc Analysis of the U.K. Money Demand in Monetary Trends and the United Kingdom by Milton Friedman and Auna J. Schwarz," Amencan Economic Review, 8 1,8-38.

Hordrick, R.J. 1978. "An Empirical Analysis of the Monetary Approach to the Determination of the Exchange Rate," In Jacob A. Frenkel and Harry G. Johnson ( eds.), of Exchaqge Rates. Reading, MS: Addison-Wesley.

Johansen, S. and K. Juselius, 1990. "Maximum Likelihood Estimation and Inference on Cointegration - With Applications to the Demand for Money, Oxford Wletin of Economics

d Statistics, 52, 169-2 10.

Johansen, S. 198 8. " S tatistical Analysis of Cointegration Vectors," Journal of Economic w i c s and ControL 12,23 1-254.

Johnson, H.G. 1971. The Keynesian Revolution and the Monetarist Counterrevolution, American Economic Review4 61, No.2,l-14. Reprinted in Ch.7 In H.G. Johnson, Econornics

d Society, Chicago: University of Chicago Press 1975.

Johnson, H.G. 1975. "World Inflation and the International Monetary Systems," The Three s Review, No. 107,3-22.

Kapur, 1. 199 1 .Ghana-Adjment and Growth. 1 983- 1 99 1, Occasional Paper (International Monetary Fund) N0.86.

Magee, S.P. 1976. "The EmpuiML Evidence on the Monetary Approach to the Balance of Payrnents and Exchange Rates," American Economic Associatiorl, . .

V01.66, No.2, 163-1 70.

McDonaid, R and M. Taylor. 1994. "The Monetary Model of Exchange Rate: Long-Run Relationships, Short-Run Dynamics and How to Beat a Random Waik," The Journal of International Monev and Finance, 13(3), 276-290.

McDonald, R and M.P. Taylor. 1991. "The Monetary Approach to the Exchange Rate: Long-Run Relationships and Coefficient Restrictions," Economic Letters, Voi.3 7, 1 79- 1 85.

McNown, R. and M. Wallace. 1989. "Cointegration Tests for Long-Run Equilibrium in the Monetary Exchange Rate Model," Economic Letters, Vo1.3 1,263-267.

McNown, R and S. Wallace. 1994. "Cointegration Tests of Monetary Exchange Rate Mode1 for Three High Inflation Economies," Journal of Monev. Credit and Banki~Vol.26, No.3, 397-410.

McNown, R and M. Wallace. 1989. "Cointegration Tests for Long-Run Equilibnum in the Monetary Exchange Rate Model," Economic Letters, 3 1,263-267.

Meese, R. A. 1986. "Testing for Bubbles in Exchange Markets: A Case of Spiralling Rates?," J

. . Vol. 94,345-373.

Meese, R and K. Rogoff. 1988. "Was it Real? The Exchange Rate-Interest DifEerentiaI Relation Over the Modem Floating-Rate Period," The J o u d of Finance Vol. XLïIX, No.4, 933-948.

Meha, Y. P. and T.C. Chiang. 1994. "Tests of Exogeneity and Causality Specincation in Monetary Models of Exchange Rate Determination," A t l a n t i c & .

Ministry of Finance and Economic Planning. 1991. Economic Recov- Promamme. Structural Adiustme nt Promanme and Related Sources of Funding.

Moosa, 1. 1994. "The Monetary Mode1 of Exchange Rate Revisted," &pIied Financid conomics, 26,279-287.

Muscatefi, V.A. and S. Hum. 1992. "Cointegration and Dynamic Time Series Models," Journal of Economic Survevs, Vo1.6, No. 1, 1-43.

Mussa, ML. 1974. "A Monetary Approach to Balance of Payments Analysis," Journal of Money, Credit and Banking, 6, N0.3~333-35 1.

Mussa, M.L. 1976. "The Exchange Rate, the Balance of Payments and Monetary and Fiscal Policy Under a Regime of Controlled Floating," Scandinavian Journal of Economics, 78, N0.2~229-248.

Nelson, C.R and C.I. Plosser. 1982. "Trends and Randorn Waks in Macroeconomic Time Senes: Some Evidence and Implications," Journal of Monetarv Economics, 10, 139- 162.

Ostedd-Lenum, M. 1992. "A Note With Quantiles of the Asymptotic Distribution of the Maximum Likelihood Cointegration Rank Test Statistics," Oxford Bulletin of Economics . . and Statistics, 54(3), 461-471.

Pagan, A. R and M.S. Wickens. 1989. "A Survey of Some Recent Econometnc Methods," conomic Journa 99,962-1025.

Phillips, P.C.B. and P. Perron. 1988. "Testing for Unit Roots in T h e Senes Regression," Biometrika, 75,335-346.

Pigou, A.C. 1920. "Some Problems of Foreign Exchmges," Economic Journa 30, No. 1 20, 460472.

Pring, M. J. 198 1. How to Fore cast In terest Rates: A Guide to Profit for Consumers, New York: McGraw-Hill Book Company.

Sach, J.D. and F. Larrain. 1993. Macroeconomics in the Global Economv, New Jersey: Prentice Hail.

Spiro, P.S. 1989. Real Interest Rates and In vestmen t and Rorrowing St ra tw New York: Quorum Books.

Sarantis, N. 1994. "The Monetary Exchange Rate Mode1 in the Long Run: An Empirical Investigation," Weltwirschaftliches-Archiv. 1 30(4), 698-7 1 1.

Foundations of Econornetric Mode11 Spanos, A. 1986. Statistical iw, Cambridge: Cambridge University Press.

Solnik, B. 1994. Intemational Investments (Third Edition), Massachusetts: Addison-Wesley Publishing Company.

. . United Nations. 1 994. International Trade S tafistics Yearbook, New York: United Nations Publications.

Van Den Berg, H. and S.C. Jayanetti. 1992. "A Novel Test of the Monetary Approach Using Black Market Exchange Rates and the Johansen-Juseh Cointegmtion Method," Economiç Letters, 41,413-41 8.

White, H. 1980. "A Heteroskedasticity-Comistent Matrix Estimator and a D k c t Test for Heteroskedasticity," Economeûica, Vo1.48, 8 17-83 8.

White, K.J. 1993. Shazam Econometrics Cornnuter Promamme User's Reference Manual Version 7.0, Toronto: McGraw Hill.

World Bank (Economic Development Institute). 1989. Successfil Development in BfDca: Case Studies of Projects. P r o m e s and Policies. Washington D.C. : Library of Congress.

World Bank. 1995. Trends in Developinp Contries, Washington D.C.: World Bank Publications.

APPLIED I W G E . lnc 1653 East Main Street

,=: Rochester. NY 14609 USA -- -- - - Phone: 71 6/48~-03OO -- -- - - F m 716/28&5989