Shadow Prices for Trade Strategy and Investment Planning in Egypt O' SWP521 John M. Page, Jr. WORLD BANK STAFF WORKING PAPERS Number 521 Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized
216
Embed
Shadow Prices for Trade Strategy - World Bank Documents
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Shadow Prices for Trade Strategyand Investment Planning in Egypt O'
SWP521John M. Page, Jr.
WORLD BANK STAFF WORKING PAPERSNumber 521
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
Pub
lic D
iscl
osur
e A
utho
rized
WORLD BANK STAFF WORKING PAPERSNumber 521
Shadow Prices for Trade Strategyand Investment Planning in Egypt
All rights reservedManufactured in the United States of America
This is a working document published informally by The World Bank. Topresent the results of research with the least possible delay, the typescript hasnot been prepared in accordance with the procedures appropriate to formalprinted texts, and The World Bank accepts no responsibility for errors. Thepublication is supplied at a token charge to defray part of the cost ofmanufacture and distribution.
The views and interpretations in this document are those of the author(s) andshould not be attributed to The World Bank, to its affiliated organizations, or toany individual acting on their behalf. Any maps used have been preparedsolely for the convenience of the readers; the denominations used and theboundaries shown do not imply, on the part of The World Bank and itsaffiliates, any judgment on the legal status of any territory or any endorsementor acceptance of such boundaries.
The full range of The World Bank publications is described in the Catalog ofWorld Bank Publications; the continuing research program of the Bank is outlinedin World Bank Research Program: Abstracts of Current Studies. Both booklets areupdated annually; the most recent edition of each is available without chargefrom the Publications Distribution Unit of the Bank in Washington or from theEuropean Office of the Bank, 66, avenue d'Iena, 75116 Paris, France.
Library of Congress Cataloging in Publication Data
Page, J. M. (John M.) 1949-Shadow prices for trade strategy and investment
planning in Egypt.
(World Bank staff working papers ; no. 521)1. Shadow prices--Egypt. 2. Egypt--Commerical policy.
3. Investments--Goverment policy--Egypt. I. Title.II. Series: World Bank staff working paper ; no. 521.HB235.E34P33 1982 338.5'2'0962 82-8594ISBN 0-8213-0009-1 AACR2
ABSTRACT
This paper presents estimates of efficiency and social accounting
prices for commodities and factors of production in Egypt which are
appropriate for the period 1979-1985. Shadow price estimates for commodities
are based on a modified input-output method which decomposes domestic supply
prices into foreign exchange and primary factor content. Particular attention
has also been given to modeling of the market for skilled and unskilled labor
in the formal sector and to determination of the shadow wage.
ACKNOWLEDGEMENTS
The paper was prepared as an element of the work program on Problems
of Trade Strategy and Investment Planning in Egypt coordinated by the Egypt
Division of the Europe, Middle East and North Africa Regional Office. I am
grateful to Vinod Dubey, Kemal Dervis, Amar Bhattacharya, Heba Handoussa, and
participants in seminars at the World Bank and the Egyptian Institute for
National Planning for helpful comments and contributions on previous drafts.
Table of Contents Page No.
Chapter 1: Introduction and Summary 1
1.1 Concepts and Assumptions 11.2 Method 81.3 The Main Results and Some Implications 15..
Chapter 2: Accounting Ratios for Commodities 27
2.1 The Valuation of Commodity Inputs 272.2 Exchange Rate Changes & Multiple Exchange Rates 432.3 Data and Assumptions 48
2.3.1 The Cost Composition of Traded Goods 492.3.2 Non-Traded Goods and Services 52
2.4 Household Income and Expenditure: Conversion Factors forConsumption and Savings 54
2.5 Conversion Ratios for Other Expenditure Categories 70
Appendix 2-1: Cost Coefficients for Commodities in Egypt 73Appendix 2-2: An Accounting Price for Cotton 99
Chapter 3: Discount Rates, The Value of Public Income, andConsumption Distribution Weightt 103
3.1 The Marginal Productivity of CaDital at Border Prices 1043.2 The Consumption Rate of Interest 1123.3 The Accounting Rate of Interest 1153.4 Relative Weights for Per Capita Consumption 118
3.5 A Possible Value for n 1223.6 The Value of Public Sector Income (v) 1253.7 Consumption Distribution Weights and the Critical
Consumption Level 139
Appendix 3-1: Rural-Urban Price Differentials in Egypt 1979/80 144
Chapter 4: Shadow Wage Rates for Skilled and Unskilled Labor 147
4.1 Labor Force, Wage and Employment Trends in Egypt 1474.2 A General Formula for the Shadow Wage 1554.3 The Shadow Wage for Skilled Labor 1594.4 Shadow Wages for Unskilled Labor 170
Appendix 4-1: Formulae for the Shadow Wage When Wages are BidUp and Test for Biases in the Estimates of Shadow Wages 192
References
List of Figures and TablesPage No.
Chapter 1
Table 1-1: Calculating the Accounting Ratio for a Commodity 11
Figure 1-1: Distribution of Accounting Ratios 14
Table 1-2: The Main Accounting Ratios 20
Chapter 2
Table 2-1: Accounting Ratios for Factors of Production in Egypt 29
Figure 2-1: Modified Input-Output Tableau for AccountingPrice Estimates 31
Table 2-2: Accounting Ratios by Category/Egypt 33
Table 2-3: Accounting Ratios for Urban Consumption 60
Table 2-4: Accounting Ratios for Rural Consumption 61
Table 2-5: Accounting Ratios for Savings of Households 66
Table 2-6: Alternative Values of the Accounting Ratio for
Workers' remittances 66
Table 2-7: Conversion Factors for Investment 72
Figure A2-1: Shadow Price Determination in SegmentedCommodity Markets 74
Table A2-2: Accounting Ratios by Category/Egypt 83
Table A2-2-1: The Elasticity of Demand for Cotton Exports 101
Table A2-2-2: Sensitivity Analysis of Marginal Export Revenue 101
Table A2-2-3: The Accounting Price for Cotton Exports 102
Chapter 3
Table 3-1: Estimates of the Marginal Productivity of* Capital ICOR Method 106
Table 3-2: Estimates of the Marginal Productivityof Capital Share Method 106
Table 3-3: Financial and Economic Rates of Return to
Some Public Sector Industrial Investments 108
Table 3-4: Rates of Return to Public Sector Industry 109
Table 3-5: Possible Values for the Consumption Rate of Interest 114
Table 3-6: Alternative Estimates of the Accounting Rate of
Interest 117
Table 3-7: Relative Weights for Marginal Changes in Urban
Incomes at Different Income Levels 120
Table 3-8: Relative Weights for Marginal Changes in Rural
Incomes at Different Levels Measured at Urban Prices 121
Table 3-9: An Estimate of n from the Egyptian Tax Tables 125
Table 3-10: The Social Value of Urban Consumption Subsidies 128
Table 3-11: The Value of Public Sector Subsidies 129
Table 3-12: The Value of Public Investment 137
Table 3-13: Best Estimates of the Value of Pubic Income
in Alternative Uses 138
Table 3-14: Consumption Distribution Weights for Urban
Households 140
Table 3-15: Consumption Distribution Weights for RuralHouseholds 141
Table A3-1: Rural and Urbar Relative Prices 1980 149
Chapter 4
Table 4-1: Labor Force in Egypt by Industry, 1960-76 152Table 4-2: Labor Force [n Egypt by Industry, 1971-79
Ages 12 to 65 153Table 4-3: Money Wages in Agriculture, Establishments and
Public Administration, 1966-76 153Table 4-4: Average Wages in Construction and Agriculture 154
Table 4-5: Sectoral Wages, 1966 and 1975 155Table 4-6: Shadow Wages for Skilled Labor Drawn from
Various Sectors of Origin 165Table 4-7: Estimates of the Proportions of Workers Drawn
from Migrant Employment Pf 168Table 4-8: Sensitivity of the Accounting Ratio for Skilled
Labor to Estimated Social Values of Income Transfer andProportion of Foreign Workers 169
Table 4-9: Distribution of Rural Household "Income Claims"
by Source 174Table 4-10: A Conversion Ratio for Agricultural Production 176Table 4-11: The Resource Cost of Unskilled Workers Changing
Occupations 180Table 4-12: Welfare Gains to an Agricultral Worker Changing
Occupations 180Table 4-13: Proportions of Workers Drawn into Formal Sector
Employment 186Table 4-14: Shadow Wage Rates for Urban Unskilled Labor 189Table 4-15: Shadow Wage Rates for Rural Unskilled Labor 190Table 4-16: Estimated Shadow Wages and Accounting Ratios
for Rural and Urban Unskilled Labor 191Figure 4-1: Impact of an Increase in Wages on the Labor Market 193Table A4-1: Estimated Biases in the Shadow Wage When Rural
Wages are Bid Up 201
Chapter 1: Introduction and Summary
1.1 Concepts and Assumptions
This paper presents estimates of shadow, or as we shall frequently
call them, accounting prices for goods and factors of production in Egypt. In
this chapter we briefly introduce the methodology which provides the framework
for our estimates and discuss the major results. Chapters 2 to 4 provide
detailed information on the data, assumptions and methods employed in our work.
Shadow prices may differ from prices observed in the market because
of non-competitive behavior, externalities, and distortions introduced by
government policy. In Egypt, as in many developing economies, policy induced
distortions are perhaps the most pervasive and at the same time the most
complex cause of divergence between market and shadow prices. In these
circumstances market signals may provide an inadequate guide for investment
planning and project appraisal, and the need for a consistent set of prices
which reflect the resource costs and social benefits of a proposed course of
action becomes apparent.
The general procedure used to estimate the shadow price of a given
resource is to determine the change in social welfare which occurs as a result
of withdrawing one unit of the good or factor from its existing employment in
the economy. This "welfare accounting" approach implies that accounting
prices depend on both the objectives of the country and on the economic
environment within which the marginal changes take place.
- 2 -
There exist in the literature two basic approaches to the social
valuation of resources. The first argues that shadow prices should be defined
solely with respect to the obj c-time-o-f-economic-elfficiency. Other objectives
are recognized but the use of policies to achieve those objectives is
generally perceived as having an economic cost valued at efficiency prices.
An alternative approach is to build into shadow price estimates the weights
attached to various national goals. Benefits accruing to different income
groups, for example, may be assigned different-values taking into account the
objective of altering the income distribution, or savings may be valued
differently from consumption in circumstances where there are constraints on
achieving the desired level of investment and growth.
The approach adopted here is to provide-estimates of accounting
f prices based both on the efficiency criterion and on a limited set of othergoals. In many instances, for example in evaluating the shadow prices of
commodities, the two sets of prices are nearly identical. There are, however,
some important parameters such as the shadow wage rate where efficiency prices
and social accounting prices may diverge significantly. In an economy like
Egypt's, where there is a long and extensive history of interventions by the
government designed to alter the distribution of income among contemporaries,
the need to have a consistent set of prices which reflect as closely as
possible the social gains and losses associated with changes in the
distribution of income is apparent. In the absence of such a set of prices,
it is impossible, for example, to arrive at a meaningful judgement concerning
the costs and benefits of the program of consumer subsidies.
If there are historical policy-imposed distortions in the economic
system, the estimation of shadow prices will vary depending on whether these
-3-
distortions are expected to be removed. In calculating "first-best" shadow
prices one has to assume that all government imposed distortions will be
eliminated except insofar as they contribute to the optimal allocation of
resources. In this sense "shadow pricing" amounts to predicting the market
prices which would prevail when policies have been changed. "Second-best"
shadow prices, on the other hand, assume that existing non-optimal policies
will remain in effect during the period of the analysis. This is equivalent
to deriving the first-order conditions for welfare maximization with market
distortions acting as constraints.
We have in general assumed that the economic environment in Egypt
will not alter radically in the next few years, and the shadow prices we
estimate reflect the taxes, subsidies and other interventions prevailing in
the period 1979-1981,. although where there is evidence of changing policies,
we have attempted to take into account the emerging trend. This should not be
interpreted as implying that we view existing policies as necessary or
permanent; indeed; as we shall indicate below, our shadow prices provide
substantial information that can be used as a basis for designing policies to
remove many of the existing price distortions.
Given these assumptions, if all goods are (potentially) tradable, the
efficiency price (or opportunity cost) of-any commodity or factor is given by
the decline in national income expressed in terms of world prices which occurs
as a result of withdrawing the resource from alternative uses. 1/
1/ There are, of course, a number of restrictive assumptions including theabsence of quantative trade restrictions and monopoly power ininternational trade which must be met, but these may be relaxed at thecost of complicating somewhat the welfare accounting process. We deal in
greater detail with these issues in Chapter 2.
The justification for this valuation, rather than using domestic prices that
reflect consumers' marginal preferences, is that as long as the ratio of
domestic market prices to world prices is constant, consumer prices are
invariant with respect to changes in the allocation of resources. In other
words, it is possible to deal with the productive sphere of the economy
without having to worry about changes in the consumption sphere. Welfare
depends only on the level of purchasing power over the world's goods and
services and can be measured by national income denominated in world prices.
Because of the possibility of trade at fiscal world prices, resources should
be allocated so as to maximize production expressed in these world prices.
In social pricing changes in the incomes of various groups in the
society become relevant. Social benefits must be expressed in terms of a
common unit of account or numeraire. Any commodity or factor may be chosen
for this purpose - in the efficiency pricing criterion discussed above, we
have implicitly chosen foreign exchange as the unit of account - but once it
is chosen all values must consistently be expressed in terms of the numeraire.
In this paper, we follow the now well established practice of valuing
all commodities and factors in terms of their impact on uncommitted foreign
exchange in the hands of the Egyptian government. 1/ Thus we retain as the
basis of our welfare accounting system the use of international prices, but we
add the additional consideration that incomes at international prices accruing
to various groups in the society may be more or less valuable than uncommitted
foreign exchange in the hands of the government. The social price of a
1/ This is the Little and Mirrlees (1974) unit of account.
-5-
commodity is, therefore, equal to the efficiency price plus an adjustment for
the distributional impact of altering the existing allocation of resources.
(See Squire and van der Tak (1976)).
Social pricing will bias project selection according to the value
judgements implicit in the adjustments made to reflect social objectives.
First, we simply attempt to establish the value of average private consumption
relative to uncommitted social income. This value depends on the uses to
which the government puts its income and on the change in the value of average
consumption resulting from these uses. In cases where the value of public
income exceeds the value assigned to average consumption, social prices
reflect the objective of increasing uncommitted public income. The accounting
prices which we estimate based on adjustments for changes in the levels of
public sector and average private sector incomes we have called "extended
efficiency prices". Secondly, we have considered the case in which welfare
weights are assigned to the incomes of various classes of private individuals.
Hence, it is not only the distribution between public sector income and
private consumption which matters, but also the distribution of private sector
income among individuals. In this system of shadow prices, it is possible for
the welfare weights assigned to some classes of individuals to exceed the
value of public sector income, and the shadow prices derived from these
weights reflect the objective of increasing the incomes of certain classes of
private individuals relative to both their contemporaries and to public sector
income. We have called the set of accounting prices based on this set of
objectives "social accounting prices".
Not all commodities are traded, however. It is the presence of
non-traded goods that constitutes the real challenge to shadow pricing.
Because the price for non-tradables is set in the local market, changes in the
-6
supply of a non-traded commodity can affect domestic consumer prices.
Moreover, as a result of market imperfections or policy induced distortions
(for example indirect taxes), the marginal value of non-traded goods to
consumers may differ from their marginal production cost. This divergence
does not create problems in the case of traded goods because their supply can-
always be increased at a fixed world price. In the case of non-tradables,
however, it is important to know whether an incremental unit of the good is
obtained by an increase in production, by depriving other users, or by a
combination of the two. Depending on which of the three cases applies, the
shadow price of such goods may be the demand price (marginal social benefit),
the supply price (marginal social cost), or may lie between the two.
There are no fully satisfactory solutions to dealing with these
complications short of trying to build a full empirical general equilibrium
model of the economy. We have generally assumed that incremental changes in
the demand for non-traded goods are met by an increase in supply, and that the
per unit cost of production is constant. The result of these two simplifying
assumptions is to allow us to approximate the shadow price of a non-traded
good by its marginal social cost of production. Moreover, because the
long-run supply curve is horizontal shadow prices are again independent of the
structure of domestic preferences. This in turn allows us to express the
marginal social cost of production of a non-traded commodity in terms of the
costs of the inputs which are needed for its production. Some of these inputs
will be traded commodities that can be valued at world prices. Others will be
non-traded inputs and finally there are, of course, primary inputs in the form
of capital, land and labor. Wherever we encounter a non-traded input in this
chain, we decompose the cost of producing it until finally we are left with
-7
only primary_fa-aco.r_ costs and traded input costs. In this way the
international prices of traded commodities determine the efficiency prices of
tradables and primary inputs which in turn determine the efficiency prices of
non-tradables. Extended efficiency prices and social accounting prices of
non-tradables differ from their efficiency prices only insofar as changes in
the output of non-tradables affect the distribution of incomes between the
public sector and various private individuals.
Our welfare accounting approach to the estimation of factor shadow
prices is essentially partial equilibrium in nature. In general equilibrium
consumption and production of most goods would be expected to change as a
result of withdrawing a unit of any resource from the rest of the economy.
Estimating these changes in a world of many commodities would be a formidable
task. One alternative is to work with a limited number of aggregate commodity
groups, such as imports, exports and non-tradables. The conditions for
aggregation, however, are very stringent and the possibility of estimating
operationally useful shadow price parameters from such an aggregate approach
is quite limited.
A more practical alternative is to investigate the activities from
which the factor is likely to be withdrawn in order to estimate the change in
national income measured in terms of the unit of account which would occur as
a result of withdrawing the factor from its alternative employment. This
procedure requires considerable knowledge of how the relevant product and
factor markets work, and to the extent that the government is directly
involved in allocating resources, it also requires an understanding of how its
decisions are likely to be made. This is the approach which we have adopted
to estimate the social opportunity cost of such factors of production as
skilled and unskilled labor and capital.
-8-
1.2 Method
If the benefits and cost of projects are to be consistently and
efficiently weighed up according to the principles outlined above, they must
be expressed in terms of a common unit of account. What we have done in the
remainder of this paper is to show the methods by which we express the
accounting prices of many commodities and factors of production in terms of
our unit of account which is uncommitted foreign exchange in the hands of the
Egyptian government. By foreign exchange we mean currency which is
convertable into any other. For conveience, we shall express foreign exchange
in terms of its Egyptian Pound equivalent, but we are still referring to
foreign exchange and could express our shadow prices in dollars or any other
foreign currency. 1/ We do not assume that all benefits and costs of a
project take the form of actual increases or decreases in the foreign exchange
reserves of the public sector. Rather we attempt to express the actual
benefits and costs in terms of equivalent value to such increases or decreases.
Our results could be expressed either as absolute accounting prices
or as the ratio of accounting prices to market prices (accounting ratios).
Although we give a few absolute accounting prices, for example, the shadow
wage and the accounting rate of interest, all of our results are expressed as
accounting ratios as these are generally more useful in giving an overall view
of the structure of price distortions in the economy.
(a) Valuing Commodities
Given our assumptions,-the social value of commodities is equal to the
value at international prices of delivering them to a representative project.
1/ In the case of multiple exchange rates it is important to choose a
consistent accounting rule for the conversion of "border pounds" into
foreign currency values. A discussion of these issues and our accounting
rules is presented in Section 2.2 below.
-9-
For tradable goods, this consists of the world price of the commodity plus the
value at international prices of the port charges, transportation and handling
required to deliver it to the user. In order to find the equivalent value at
international prices of non-tradables, given our assumption of constant
marginal costs, we decompose costs of production into tradable inputs,
non-tradable inputs and the costs of a class of inputs which we shall define
as "primary inputs", for example, labor and foreign exchange. We know the
shadow prices of the traded commodities used as inputs; further, we can break
down the non-tradable inputs into their traded, non-traded, and primary input
cost components. If we move through several rounds of analysis in this
fashion, we can reduce the social cost of any commodity to its equivalent cost
in terms of tradable commodities (foreign exchange) and primary inputs. This
is simply an application of input-output analysis. The shadow price of the
commodity is then given by the sum of the shadow value of traded inputs plus
the shadow value of the primary inputs.
The problem then reduces to estimating the shadow prices of primary
inputs. Unfortunately, the shadow prices of primary inputs themselves depend
on changes in national income (or public sector income) denominated in terms
of foreign exchange, so that they depend ultimately on the shadow prices of
commodities. The solution to this simultaneity problem is simple in
principle; all we need to do is to build a model which will simultaneously
provide the shadow prices of commodities and factors. In practice it is
difficult to implement in a world of many goods and factors.
The main device which we have used is first to express the cost of
any commodity as the sum of its direct plus indirect primary input costs (at
market prices). Because all production functions are linear, the total
(direct plus indirect) primary input content of all commodities may be
determined by the solution to a large number of simultaneous linear equations
- 10 -
(see Chapter 2). We then make rough estimates of the accounting-ratios for
the primary inputs, and multiply the total primary input costs by these
accounting ratios. This gives us corresponding accounting ratios for
commodities. We then use these preliminary accounting ratios for commodities
to arrive at better estimates of the accounting ratios for primary inputs and
by iteration arrive at a mutually consistant. set of accounting ratios for
everything.
The essentials of the method are illustrated in Table 1-1. Consider
a project which uses I.E 100 worth of packaging materials which are imported as
an input. The customs duty on these is 30 percent, assuming that no special
rebate is allowed, and the port handling, transportation, and handling margins
are as listed in the table. By our input-output method, we may convert these
direct inputs into the direct plus indirect primary input cost coefficients
listed in column (2). We then multiply the total primary input cost
coefficients by their respective accounting ratios and arrive at a weighted
average accounting ratio of .811. The total cost of packaging materials to
the project is therefore LE 81 at border prices.
Several points may be noted. The accounting ratio for foreign
exchange is simply 1.00 since this is our unit of account. The accounting
ratios for taxes, and subsidies are zero, since these are simply transfers
within the public sector. The accounting ratios for capital inputs reflect
their foreign exchange costs; those for labor reflect the estimated change in
national income evaluated at border prices which would occur as a result of
diverting a unit of labor from its existing allocation, expressed as a
percentage of the market wage.
- 11 -
Once we have the accounting ratios for a large number of goods and
services we need not always increase the size of our input-output table to
find an accounting ratio for a single new commodity. If the accounting ratios
Table 1-1: Calculating the Accounting Ratio for a Commodity
Cost Component Proportion of Accounting Accounting Cost
The main results of our analysis are summarized in Table 1-2 and in
Figure 1-1. Table 1-2 presents accounting ratios for primary inputs and for a
number of categories of expenditure, as well as the values of a number of
distributional parameters which are intermediate inputs into the estimation of
social accounting prices. Figure 1-1 summarizes the frequency distributions
of accounting ratios for the'130 commodities included in the study and for
important subsets of these commodities.
The most obvious point which'emerges is that the median accounting
ratio is close to one. Those familiar with other studies of (border)
accounting prices or effective protection in developing countries may find
this result striking. In most countries, protection of a wide range of
tradable consumer and intermediate goods results in accounting ratios for.
tradable commodities which are below unity, often by a wide margin. The
median of our sample of 94 traded commodities is .97, and fully 43 percent of
the accounting ratios for tradables exceed one. Since non-traded goods are
linear combinations of tradables and primary inputs it is not very surprising
to find that their accounting ratios are also close to and often exceed unity.
What is the source of this shadow pricing phenomenon in Egypt? Since
the opening of the Egyptian economy in 1973, the government has tried to
insulate certain consumer goods and producer intermediates from changes in
world prices and in the nominal exchange rate via a system of price controls
and subsidies. Among these commodities are domestically consumed petroleum
products, basic agricultural staples, including wheat and other food grains,
oils, sugar and meat, locally produced and imported intermediate products such
as steel reinforcing rods, cement, and other construction materials, and
- 16 -
consumer products produced by the public enterprises including shoes,
clothing, and cotton cloth. Various rationing schemes have been implemented
to control excess demand in the markets for these commodities and prices have
not been allowed to increase in line with international prices.
At the same time, agricultural policy has permitted the coexistance
of implicit taxes via fixed producer prices set at levels below the import and
export parity prices of agricultural commodities combined with subsidies on
major agricultural inputs. One consequence of this structure of subsidies and
taxes has been to make the accounting ratios for a large number of
agricultural inputs and outputs greater than one. A possible exception to
this rule is the accounting ratio for extra long staple cotton. Our analysis
in Chapter Appendix 2-2 indicates that the farmgate price of extra long staple
cotton, a product over which Egypt exercises monopoly power, may lie above its
marginal export revenue, although our best estimate of the accounting ratio is
about 1.17.
The overall impact of the system of taxes and subsidies is apparent
from Figure 1-1. The median of our full distribution of 130 commodities is
.99. If this is typical of the central tendency of the distribution of
accounting ratios for all commodities in Egypt (including those whose
accounting ratios we have failed to estimate) it indicates that, on average,
there is little overvaluation of the Egyptian currency resulting from the
structure of nominal protection. In other words, the shadow exchange rate for
the Egyptian economy lies very close to the official exchange rate. This
result is a consequence of the existing structure of taxes and subsidies in
Egypt. Since the shadow exchange rate in the Little-Mirrlees sense is simply
a'summary measure of the relationship between world prices and domestic
prices, our finding that it is close to the official rate (that is that the
standard conversion factor is close to one) indicates that the distortionary
effects of import duties, export taxes and consumer subsidies approximately
cancel out in the aggregate. If the existing structure of protection were to
be eliminated and domestic prices were permitted to adjust to their border
price equivalents, there would be about as many domestic prices that would
rise as there would be prices that would decline in the absence of any
adjustment in the nominal exchange rate. It is likely then that there would
be very little need for an exchange rate adjustment to compensate for the
change in the pattern of taxes and subsidies. This result contrasts with the
more usual estimates of standard conversion factors which are less than
unity. In those cases prior to any nominal exchange rate adjustment the
prices of most goods would decline to their international equivalents. The
nominal exchange rate would then increase to restore equilibrium in the goods
market.
It is important to note here that the exchange rate adjustment
discussed with respect to changes in the structure of protection is a
comparative static one, reflecting changes in the levels of tariffs, subsidies
and price controls. Our accounting ratios fail to provide any information on
the equally important dynamic question of possible exchange rate adjustments
in response to long-term changes in the availability of external resources.
Because Egypt derives a significant portion of it foreign exchange from such
"exogenous" sources as rents paid on petroleum exports and the Suez Canal as
well as from workers' remittances, changes in the availability of these
resources, for example arising from exhaustion of petroleum reserves, may
- 18 -
result in the need for an adjustment of the exchange rate. Any resulting
change in the relative prices of traded and non-traded goods would need to be
reflected in our shadow pricing system. This problem of possible shifts in
long-term resource availabilities is considered in a paper which will appear
as a companion to this study. 1/
Focusing on the median, however, tends to obscure important sectoral
variations in the structure of relative accounting and market prices in
Egypt. There is great variability in the accounting ratios. The standard
deviation of the scatter of 130 commodities is 1.70 and exceeds the mean by
ten percent. Clearly, the process of administrative price determination in
Egypt has resulted in extreme variability in the relationship between market
prices and opportunity costs. Thus, any summary measure of the pattern of
price distinctions such as the standard conversion factor or the shadow
exchange rate marks important sectoral variations in the relationship between
accounting prices and market.prices.
These general observations are borne out by a more detailed
examination of certain subsets of commodities. The median value for exports
is .98 which exceeds slightly that for importables of .96. In both cases, the
standard deviation is less than the mean, indicating that the distributions
are somewhat more compact than the overall distribution.
The closeness of the median accounting ratios for importables and
exportables indicates that the expected gains from shifting resources within
the traded goods sector from import competing to export production are likely
to be small. Reductions in the output of one sector carry a social cost which
is almost precisely offset by the social gains from the increase in the other
1/ See K. Dervis, R. Martin, S. van Wijnbergen: An Optimal Growth Model for
Egypt: Issues in Investment Allocation, Shadow Pricing and Oil
Extraction. World Bank, (forthcoming).
- 19 -
sector' s output. The median values do mask substantial variations in
individual accounting ratios, however, and therefore fail to provide
sufficiently accurate measures of the opportunity costs of individual
commodities. Proposed investments in the traded goods sector should be
appraised individually. Summary conversion factors are inadequate guides to
the social costs and benefits of any particular investment. Non-traded goods
and services are more closely grouped (coefficient of variation .54) which
reflects the fact that they are the weighted averages of a large number of
items. Both the mean and the median of the distribution of non-traded goods,
however, exceed one by a substantial margin (1.33 and 1.21, respectively).
When we recall that the accounting ratio for a non-traded good is the ratio of
its marginal social cost of production to the domestic market price, the
impact of the system of subsidies on intermediate goods and on the
relationship between market and social profitability becomes apparent. Those
activities which intensively use such non-traded goods and services as
electricity, construction, and road and rail transport receive substantial
implicit subsidies relative to the opportunity cost of providing those
inputs. If the accounting ratio applicable to their output is close to the
median value of one, it is quite likely that profits at market prices will
exceed social profits. Moreover, the fact that private costs are below social
costs for these items may encourage excessive substitution of these inputs for
others relative to the optimum input proportions at social accounting prices. 1/
1/ It is perhaps useful to re-emphasize here that our accounting ratios fornon-traded goods are supply prices and hence are appropriate to evaluatingthe social cost of an input. They do not provide the accounting valueappropriate to the output of a project producing the non-tradable. Thisis set by the demand price for-the commodity converted into equivalentvalue in terms of the unit of account
117 23 Port and Harbor Charges 1.315118 24 Printing and Publishing 1.236119 25 Rail Transport (passengers) MC 1.199120 26 Rail Transport (goods) MC 1.213
121 27 Rail Transport (price) MC 0.000
122 28 Retail Distribution 0.711123 29 Road Maintenance 1.423
124 30 Road Transport (passenger) 1.270125 31 Road Transport (goods) 1.365
126 32 Services NES 1.627
127 33 Telecommunications 0.972128 34 Transport (wto png) 1.343
129 35 Vehicle Repair 0.909
130 36 Wholesale Distribution 0.798
The 130 commodities in Table 2-2 are divided into six major groups -
tradable intermediates, agricultural inputs and outputs, petroleum products,
tradable urban consumer goods, tradable rural consumer goods, and
non-tradables. Our coverage of commodities includes all of the major SITC
three digit categories in Egypt's external trade statistics, major consumer
goods, and all of the non-tradable sectors represented in the 1977
input-output table. Thus the frequency distribution of accounting ratios in
Table 2-2 is quite representative of the overall relationship between
accounting and market prices in the Egyptian economy.
- 37 -
A summary measure of the "typical" accounting ratio is the median of
the distribution of all 130 commodities which is .993. The mean is 1.549
indicating the effect of a number of commodities with extremely high ratios of
the accounting price to the market price. There are relatively few
commodities with low accounting ratios in our distribution, and these are
consumer goods which are typically heavily taxed, tobacco, coffee, tea and
alcohol. Because of the heavy subsidization of many consumer items, the
median accounting ratio for tradable consumer goods is .970. The median of
the distribution of petroleum products is 5.88 indicating the magnitude of the
subsidies to petroleum products. Subsidies and production taxes in the
agricultural sector are similarly indicated by the median of the distribution
of agricultural inputs and outputs of 1.28. The system of subsidies also
works its way through the structure of production and exerts a major influence
on the social marginal cost of non-traded commodities. Non-traded goods in
our sample have a median accounting ratio of 1.21, thus for the typical
non-traded good the accounting price exceeds its market price by approximately
21 percent.
Overall the picture which emerges is of a distribution of accounting
ratios characterized by high variance and a large incidence of values
exceeding unity. There are few heavily protected sectors of the Egyptian
economy, and in contrast to many developing countries there are a substantial
number of activities in which pricing policy has suppressed the domestic
market price below the shadow price. The most notable of these sectors is
petroleum, but accounting prices exceeding unity are also found in agriculture
and among subsidized urban consumer goods.
- 38 -
Before turning to a detailed discussion of data and the methods used
to establish the input structures of the commodities listed in Table 2-2, it
may be useful to review briefly a number of general issues which affect the
appropriateness of our methodology. These are the degree of tradability,
increasing returns to scale, variable relationships between international and
domestic prices, and inelastic world supply and demand. These topics also
introduce the important issues of the time perspective of the analysis and the
degree to which individual project analysts should undertake supplementary
shadow price estimates in addition to those presented here.
(a) The Degree of Tradability
A crucial underpinning of the border price rule on which our system
of accounting prices is based is that potentially tradable goods are in fact
traded at the margin. In the absence of that assumption -- s.ay for example
because of fixed import quotas or heterogeneity of products -- changes in the
domestic supply of or demand for potentially tradable commodities may result
in changes in their domestic price with no effect on the trade balance. In
such circumstances accounting ratios based on the relationship between border
and domestic prices will not provide a correct estimate of marginal welfare
change and, hence, of the marginal social cost/benefit associated with the
commodity. The cost-benefit analyst has essentially two options in dealing
with commodities for which the degree of tradability is uncertain. The first
is to treat all potentially tradable commodities as traded on the rationale
that it is likely that for minor changes in final demand imposed by the
project, quotas or other quantative interventions will be adjusted upward to
reflect the increased demand. The alternative is to treat such commodities as
non-traded goods. It is then possible to analyze their social costs of
- 39 -
production in the manner described above, or in the case of commodities which
are in inelastic domestic supply to estimate the changes in producer's and
consumer's surplus which occur as a consequence of a marginal change in supply
or demand. Clearly the second alternative involves substantially more work
for the project analyst and introduces a greater margin for error. Therefore
most cost benefit studies have, wherever possible, treated tradables as fully
traded. We shall follow that approach here, but with several important
exceptions. These are primarily in the area of subsized and rationed consumer
goods, notably textiles and clothing, where the level of product quality is
substantially lower than the minimum standard in international trade. Thus
although these items fall into a category which is potentially tradable, they
are in fact non-traded goods, and we shall detail the assumptions made
concerning their social cost in the notes on specific industries below.
(b) Constant Costs vs Variable Returns to Scale
The input-output method outlined above implicitly assumes that the
separate items making up the marginal social cost of a commodity vary linearly
with the quantity of the good demanded. If this were not so, the A and P
matrices would vary with the scale of demand. This is not precisely the same
thing as assuming constancy of average costs, since the accounting prices are
applicable only to marginal changes in supply or demand. Where large changes
in the availability of an item are contemplated the project analyst must, as
always, carefully consider the possible changes in social costs associated
with non-marginal changes, and the accounting ratios presented here would not
in general be applicable. Otherwise the constancy of the A and P matrices
implies constancy of marginal social costs. Allowances for variable returns
to scale may be made by allowing variations in the a.. and p . overii Si
- 40 -
different intervals of output. In practice this can be done only to a limited
extent. Moreover, given our limited knowledge of returns to scale in the
non-traded goods sectors of the Egyptian economy, any such adjustments would
be largely arbitrary, and we have therefore chosen not to make them. The
marginal social cost coefficients in our input-output structure represent the
best information we have on the cost composition of those activities.
(c) Variable Relationships Between International and Domestic Prices
Wherever domestic prices are admihistratively set, as for example is
the case with fixed producer and consumer prices, the relationship between
domestic and international prices may vary over time with changes in either
price. 1/ This has rather vexing consequences for accounting ratios based
upon the relationship between the market and shadow price of a commodity taken
at one particular moment. Although the appropriate accounting price remains
the border price, when the ratio of prices changes, the accounting ratio must
also change. In the case of agricultural commodities which exhibit
considerable year to year variations in international prices, the presence of
fixed domestic prices could imply almost continuous updating of accounting
ratios. In general, the economist estimating a set of shadow prices must use
his judgement concerning the medium term relationship beteen domestic and
foreign prices. In the Egyptian context whete there have been substantial
changes in domestic prices of price controlled traded commodities since 1979
l/ Obviously the presence of quantative restrictions is another example of a
variable relationship between world and domestic prices, since changes in
domestic demand or supply will alter the quota premium of the commodity.
- 41 -
we have in general focused on the relationship between these domestic prices
which, given past practice, may be expected to prevail for several years and
the "normal" world price as represented by the average of the past several
years prices. These accounting ratios should remain relatively stable over
the next few years, but major movements in either price would require revision
of the input-output value coefficients and the accounting ratios.
(d) Inelastic World Supply or Demand
Cases in which a country faces less than perfectly elastic foreign
supply or demand curves for a commodity pose a slight complication in the
estimation of accounting ratios, but can be allowed for within the general
approach by adjusting the input coefficients to reflect the marginal export
revenue or marginal import cost of the commodity. Social pricing exercises
will also require that the change in producer and consumer incomes resulting
from the price change be evaluated. For Egypt the major commodity market in
which the quantity exported could have a significant impact on the world price
is that for long staple cotton. Thus we devote some attention to working out
the plausible value of the marginal export revenue from cotton.
It should be clear from the foregoing discussion that the general
system of estimating shadow prices is quite flexible with regard to its
ability to accommodate various institutional and policy constraints which
affect the relationship between market and social prices.- The implementation
of these procedures, however, in order to calculate a basic set of shadow
prices for an economy such as Egypt, is a fairly laborious and time consuming
1/ Two levels of aggregation are represented by the "trade data approach" ofSquire and van der Tak (1975) and the input-output approach advocated bySchohl (1979).
- 42 -
task. This is because the construction of the A and P matrices requires
detailed analyses of the markets for commodities, a project which can use
almost unlimited amounts of information. In view of the alternative methods
available for estimating more aggregated national parameters in cost-benefit
studies what are the advantages of a more detailed approach? From the point
of view of the present study we would argue that there are three important
gains from taking the greater time required for a highly detailed shadow
pricing exercise:
(i) It requires a careful examination of the workings of many commodity
markets and industries and thus generates valuable information about
the economic factors underlying the present situation. This is of
particular value in an economy like Egypt where administered pricing
decisions impinge at both the producer and consumer margins in the
markets for many commodities. Our approach allows us to trace the
impact of these interventions through a number of interconnected
markets.
(ii) Because we wish to apply our accounting prices to an examination of
comparative advantage and international competitiveness in several
Egyptian industries, quite a wide array of accounting prices is
required, and it is convenient to estimate them within the context of
a general shadow pricing exercise.
(iii) The work, once completed, can be updated relatively easily to reflect
changes in the policy environment. Once the impact of a policy
change on the coefficients of the A and P matrices has been worked
out, it is simple to trace the impact of these changes on the full
set of accounting ratios.
- 43 -
Thus we have attempted to work with as wide a range of items as
possible within the resources available. By distinguishing between various
types of commodities - e.g. types of transport, agricultural products,
petroleum products, etc. - we are able to consider in detail the ways in which
government pricing and taxation policies are affecting the incentive structure
of the economy relative to what might be desirable in view of the structure of
social costs and benefits.
2.2 Exchange Rate Changes and Multiple Exchange Rates
Recall that we have chosen to express all values of commodities and
factors of production in terms of constant 1979 "border Egyptian Pounds" -
that is Egyptian pounds freely convertible at the average nominal exchange
rate prevailing in 1979. If one wished to switch from border pounds to
dollars as the unit of account it is simple to divide the accounting value by
the nominal exchange rate. Either unit of account provides a consistent
welfare index differing only by a scalar value.
Since 1979, however, two changes have occurred in the exchange rate
regime which must be addressed. First, there has been a marked widening in
the exchange rates offered in each of the three segments of the Egyptian
foreign exchange market - the "central bank pool", the "commercial bank pool",
and the "free" market financing own exchange imports. In 1979, transactions
in the commercial bank and central bank pools both took place at a rate of
.70 LE to the dollar, and those on the "free" market took place at rates not
exceeding .78. By 1981, although the central bank rate remained at .70, the
commercial bank rate had increased to .84 and the parallel market rate had
increased to around 1.00. Hence, the weighted average nominal exchange rate
has increased to .80. Thus, the period from 1979 to 1981 was marked by a
shift from an essentially unified to a multiple exchange rate regime and by a
- 44 -
nominal depreciation of the Egyptian pound. How do these two actions affect
our estimates of accounting prices and accounting ratios?
Let us first consider the question of a nominal devaluation. The
change in the exchange rate will have no impact on the world price of tradable
commodities (denominated in terms of foreign currency) for which Egypt is a
price taker. Domestic prices will rise by the amount of the devaluation but
relative traded goods prices will be unaffected. Accounting ratios for traded
goods will also be unaffected. To see this, consider the case of a tradable
commodity which is subject to an ad valorem tariff = (l+t). The accounting
price denominated in dollars is P and in terms of border pound is P R,
where R is the nominal average rate of exchange. The domestic price is P R T
and the accounting ratio is therefore:
f. = P R/P RT = 1/T. (3)
The level of the nominal exchange rate is irrelevant to the shadow price of
the traded commodity and to its accounting ratio. It nevertheless remains
true that because a nominal value has changed (in this case R) we have to
deflate to constant 1979 prices if we wish to compare real resource costs in
the two periods. If our unit of account is dollars and dollar denominated
world prices are unchanged this may be accomplished simply by dividing border
pounds in each period by the nominal average exchange rate prevailing in that
period. If we wish to retain border pounds of constant purchasing power as
the unit of account then we must deflate the nominal border pounds in each
period by a discount factor equal to Dt = R79 / Rt;(t = 79,.. .N).
-45 -
Either approach provides a consistent index of the real resource cost of a
traded commodity. 1/
A similar argument applies to the accounting prices and accounting
ratios applicable to non-traded goods, as long as the devaluation is nominal
and not real. A nominal devaluation implies that non-traded goods prices rise
in the same proportion as the adjustment in the exchange rate and that
therefore the relative price of traded and non-traded goods remains constant.
With constant relative prices, the structure of production is invariant to the
change in the exchange rate; hence the input-output structure on which
non-traded goods shadow prices are based remains unchanged as do the shadow
price and accounting ratio.
If the devaluation is real, in the sense that the relative prices of
tradable and non-tradable goods change after the devaluation, the effect of
the exchange rate adjustment is more complex. The relative price change will
1/ To illustrate this consider the following simple example. A project uses
100 units of an input in both 1979 and 1981 with a constant world price of
US $1.00. The commodity is subject to a tariff of 10 percent ad valorem.
The effect of the exchange rate adjustment may be seen from the following
table:
Year Units Pw Accounting Value R Value in Domestic AccountingUS $ Border LE Value Ratio
1979 100 1 100 .70 70 77 .909
1981 100 1 100 .80 80 88 .909
It is clear from the table that the real resource cost of the input -- its
accounting value in US dollars is unchanged between the two periods, as is
the accounting ratio. In moving from domestic market values to accounting
values one would first multiply by the accounting ratio to arrive at
current border pounds. To arrive at 1979 US dollar values one would
divide by the nominal exchange rate (R). To value commodities in border
pounds of constant purchasing power one would multiply the 1981 current
border value by the discount factor 70/80 = .875. Both techniques yield
the correct answer that the real resource cost of the input is unchanged
between the two periods.
- 46 -
alter the input-output coefficients on which the shadow pricing system is
based and will also change the estimates of the marginal social products of
labor and capital. A real exchange rate adjustment therefore alters the
accounting ratios applicable to non-tradables and, hence, affects the overall
distribution of accounting ratios in the economy. Predicting the net effect
of this adjustment on the accounting price system is quite difficult. 1/ In
the case of Egypt, we do not believe that the adjustment in the exchange rate
between 1979 and 1981 reflects a real exchange rate adjustment. Indeed,
increases in non-traded goods prices may have led the exchange rate,
suggesting that the nominal change from .70 to .82 reflects adjustment in the
foreign exchange market to maintain constancy of the real exchange rate.
The shift from a unified exchange rate regime to a multiple exchange
rates poses somewhat different problems. In Egypt access to various elements
of the foreign exchange market is determined by the ownership of the
enterprise as well as by the type of good. Thus, it is possible for a single
commodity to be imported at three different rates of exchange depending on the
final user or importer. The exchange rate system, therefore, operates like a
system of indirect taxes and subsidies. Enterprises with access to the
central bank pool receive an implicit subsidy on imports entering at the rate
of .70, but pay an implicit tax on exports receiving the same rate, relative
to the commercial bank and parallel market rates. A similar, although
converse argument, applies to firms importing and exporting exclusively at the
parallel rate. Differences in the exchange rates represent transfers between
1/ Very little work has been done on predicting the impact of a real exchangerate adjustment on a system of shadow prices. See Little and Mirrlees
(1976) and Ray (1982).
- 47 -
suppliers and users of foreign exchange, rather than variations in the real
resource costs of commodities supplied to various categories of users.
If the exchange rates were specific to commodities it would be
possible to handle the implicit tax/subsidy element of the multiple exchange
rate system in a symetric fashion to that applied to trade taxes and
subsidies. One would choose one exchange rate as a base and show deviations
in the market price border price relationship due to different exchange rates
as taxes or subsidies. Because in Egypt the exchange rate may differ for
differing commodity-user pairings we have adopted a slightly different
approach. We choose as a base border Egyptian pounds convertable at a rate of
.70 in 1981. (Note that this is equivalent to constant price 1979 border
Egyptian pounds and was, therefore, chosen for convenience.) We can then
specify two "exchange rate accounting ratios", EAR's, which allow us to
convert from border LE at the commercial or parallel rate to border LE at the
central bank rate. To find the accounting value of a traded commodity whose
price is denominated in terms of domestic currency in 1981, we multiply by
both the commodity accounting ratio and the exchange rate accounting ratio to
arrive at the resource cost in terms of border pounds convertible at the
central bank rate. This value in turn may be converted into dollars at the
central bank rate if desired.
The purpose of making this adjustment is very similar to the
rationale for expressing values in terms of constant purchasing power. We
wish to reflect the ultimate impact on the trade balance of a change in
expenditure at domestic prices. Consider the commodity discussed above with
world price P , tariff rate T and accounting ratio f. The domestic price
is variously P, = P RT < P= P RcT - P. P R f , depending1 w 2 w 3 w
- 48 -
on the exchange market to which the purchaser has access, where R, R and
R are the exchange rates in the central bank, commercial bank, and free
market pools respectively. Since the world price is constant, border pound
prices must also reflect the constancy of the opportunity cost of the good.
Define the exchange rate accounting ratios EAR = R/Rc and EAR =
fR/Rf. The accounting value of the commodity in border pounds (central bank
rate) may be found by:
P f = P f EAR = P f EAR = P RT(1/T) = P R T(1/T)(R/Rc) = (4)lj 2i 3j w
P R Tf(l/T)(R/R ),w
which obviously differs only by a scalar, R, from the world price P . Hence
the effective accounting ratio for any commodity-user pair consists of the
product of the commodity accounting ratio f. and the exchange rate
accounting ratio EAR (i = c,f). The exchange rate accounting ratios are
central bank rate, 1.00, commercial bank rate, .70/.84 = .833, and parallel
market rate, .70/1.00 = .70. Note that when the exchange rate accounting
ratios are applied there is no need for a further adjustment to arrive at 1979
border pound since the exchange rate is equal in both periods. 1/
2.3 Data and Assumptions
Estimation of accounting ratios by the methods described above
requires a substantial amount of information on the composition of commodity
and factor inputs used in the production and supply of goods and services. We
1/ For the project analyst whose accounts are given mainly in US dollars analternative approach is to convert the central bank border price values of
non-traded goods and labor given in this paper to dollar values using therate of .70. Either approach provides a consistent index of social valueand the choice is dictated largely by convenience.
- 49 -
have drawn our estimates of input-output coefficients from several statistical
sources of varying reliability. Inevitably some of the data are not as recent
as one would desire, and given the major structural changes in the Egyptian
economy in the past several years, this represents a limitation which should
be corrected as additional information becomes available.
The cost composition breakdowns for goods and services contained in
Table 2-2 appear in Appendix 2-1. Here we shall focus on the general sources
of data employed and highlight major assumptions and methods of deriving the
input coefficients. Two major references are available on the structure of
production in Egypt. The first and most comprehensive is the 1977
input-output table. The second is the production accounts of public sector
enterprises for which the last year available is 1979. Transport and
distribution margins may be imputed from both of these sources for several
categories of traded goods, and they provide the best information available on
the cost structure of non-traded activities. In general we have used the
public sector production accounts as our principal source supplemented by the
input-output table.
The accounting ratios estimated in this study, except in the case of
items entering into rural or urban consumption, are applicable to producer
prices. These are more appropriate for empirical studies of industrial
comparative advantage and have the additional advantage of requiring fewer
assumptions to be made regarding distribution margins.
2.3.1 The Cost Composition of Traded Goods
Cost breakdowns for more than eighty tradable goods are presented in
the Appendix tables. These goods fall into two categories, consumer goods
used in estimating the consumption conversion ratio and producer goods which
- 50 -
are significant inputs into domestic Egyptian industries. Data for the cost
analysis were drawn from published sources and from some unpublished
Government of Egypt memoranda.
Wherever possible an attempt was made to determine the cost
composition of an importable by direct comparison of cif, wholesale and retail
prices. In most instances, however, aggregation problems made direct
comparison of unit prices uncertain, and in the case of some producers goods
market prices were unavailable. Thus for approximately half of the traded
goods analyzed a border price estimate was derived from the external trade
statistics. Estimated duty rates, port handling charges, transport costs, and
distributor's margins were added to the border price to arrive at an estimate
of the final delivered costs. In other instances, principally for
non-agricultural consumer goods, it was more convenient to work backward from
the market price.
Duty rates for 1979/80 were compiled on the basis of unpublished data
provided by the Department of Customs and Excise. The compilations included
estimates of both statutory rates and actual collections. Where substantial
differences existed between the two figures, the ad valorem equivalent of the
actual collections was used as the relevant rate. If several duty rates were
applicable to items within a single category of tradable goods -- e.g.,
industrial chemicals -- a weighted average duty was estimated for the category
using the 1979 levels of imports.
Information on excise and other indirect taxes and on price
differentials charged by public sector distribution agencies were taken from
the public sector accounts and several additional unpublished sources. 1/
1/ Principally el Edel (1980).
- 51 -
For traded agricultural commodities transport and distribution margins and the
implicit subsidy for urban consumer goods were taken from two recent
agricultural price policy studies carried out by the Bank. 1/
Port handling costs were estimated from per ton handling charges at
Alexandria Harbour. Where these data did not exist we attempted to impute
"typical!' handling margins from comparable categories of goods.
Road and rail transport costs were available from data provided in
the Egypt National Transport Study (Netherlands Engineering Consultants
(1981)). The various volumes of this report provided estimates of ton/mile
charges for major commodity groups by both rail and lorry.
Perhaps the weakest link in our analysis of the cost composition of
tradables is in the distribution margins imputed on the basis of data from the
input-output table and the public sector accounts. In the absence of a survey
of distributive trades, these represent the best information available; but
the estimates could be considerably improved. Average retail and wholesale
margins derived from these sources were 19.0 and 16.8 percent respectively
which seem a bit high by comparison with other countries, especially with
respect to Egypt's relatively compact spacial distribution. We have
nevertheless employed these margins in the absence of alternative information.
There were several tradable activities which required special
consideration. Notes on these specific industries are found in Appendix 2-1.
1/ Cuddihy (1980) and World Bank (1981a)
- 52 -
2.2.2 Non-Traded Goods and Services
Cost breakdowns for forty non-traded goods and services are presented
in the Appendix tables. These are the major activities which are either
(i) important inputs into industrial production or (ii) major components of
rural and urban consumption. The principal sources of information on the
non-traded sector were the 1977 input-output table and the public sector
production accounts. These were supplemented by sector studies in
transportation and limited personal interviews. Although none of the sources
offered exhaustive information on all activities the estimates derived from
them should prove sufficiently accurate for the estimation of accounting
prices. In the remainder of this chapter we deal generally with the problems
encountered in estimating the input-output coefficients for non-traded
industries and direct specific remarks to industries where the estimates are
unusual or uncertain.
The cost coefficients for most non-traded activities are based on the
long run average cost of production. It was assumed that the industries
operated under conditions of constant marginal costs over the range of output
involved in the estimates. Hence, average Leontief type production
coefficients are appropriate for the evaluation of marginal social costs of
production. 1/ An exception to this general rule was made in the case of
railway transportation, for which evidence indicated that economies of scale
were significant, and that 1979-80 market prices differed from marginal costs.
1/ An alternative approach would have been to assume a plausible value forthe divergence between marginal and average costs for industries in which
economies of scale were believed to be significant. Data on potentialeconomies of scale in Egyptian industries are unavailable, however.
Rather than use estimates derived from data on other countries it was felt
appropriate to assume generally constant costs.
- 53 -
By assuming all costs to be variable, possible benefits from
increased utilization of industrial capacity resulting from project induced
increases in demand are ignored. Excess capacity in Egypt occurred most
frequently in those import substitution industries which are treated here as
tradable goods. Non-tradable activities which provide direct inputs into the
manufacturing sector are generally services, and output may reasonably be
assumed to keep pace with increases in demand. Thus, as a rule, the use of
long run average costs should provide accurate estimates of the marginal
social cost of non-traded goods. An exception to the general procedure was
made in the case of certain quota restricted inputs into Egyptian
manufacturing. The assumptions made regarding the marginal social cost of
these industries are dealt with in the notes on specific activities.
The Allocation of Wages and Salaries
The distribution of labor costs between payments to skilled labor and
payments to unskilled labor has important implications for the final estimates
of resource costs, since the opportunity costs of skilled and unskilled
workers differ significantly. The division between skilled and unskilled
labor is in fact a division between labor which is assumed to be fully
employed in the economy and labor which is either unemployed or
underemployed. Skilled labor is labor which is assumed to be engaged in
occupations where the market wage represents the private opportunity cost of
the worker. The distinction between skilled and unskilled workers requires
that the latter category include all workers whose jobs require limited or no
on-the-job training. Such "semi-skilled" occupations as machine paced
operatives or drivers ought to be counted as unskilled labor, since these
occupations can be readily filled with limited training.
- 54 -
The nature of the data provided on non-traded activities made a high
degree of approximation and some arbitrary classifications necessary in making
the allocation of labor costs. For approximately half of the non-traded
activities examined an estimated breakdown between payments to skilled and
unskilled labor was available. 1/ For the remaining activities the principle
sources of data on wage payments were the Public Sector Enterprise Accounts
and the ILO (1981) labor force study. The public sector accounts provided
information on the distribution of wages and salaries between production and
other (technical, managerial, and clerical) workers. In cases where this was
the only information available, all production workers were allotted to the
unskilled category while other employees were considered skilled. Where
possible, data from the public enterprise accounts were supplemented by
reference to the ILO report. All unskilled labor in industrial non-tradable
activities was assumed to be urban unskilled labor. Although this assumption
is not strictly correct for such activities as road transportation and
distributive trades where rural enterprises are important components of the
industry, it was impossible to estimate the relevant proportions of rural and
urban workers in these activities.
The Treatment of Indirect Capital Inputs
To estimate the indirect inputs of capital into a project it is
necessary to know something about the size and composition of the capital
stocks of industries supplying inputs into the project under evaluation. It
is also necessary to convert capital inputs whose economic lives extend over
more than a single period into current inputs without distorting the resulting
estimates of accounting prices.
1/ These were industries covered by the author's personal interviews or by
other firm level returns.
- 55 -
The approach adopted in this study was to identify three primary
input categories of capital -- buildings, machinery, and vehicles. Estimates
of the annual inputs of capital services provided by each of these three
categories of assets were computed for each non-traded activity. The
procedure for estimating the annual input of capital services at market prices
was quite simple. An annuity factor based on the marginal product of capital
and an assumed economic life of the capital input was applied to the estimated
value of the capital stock of the asset. This annuity gave the annual
equivalent of the depreciation and return to capital at market prices, and was
entered as a direct primary factor cost of the activity employing the capital
input. In this manner, cost coefficients for capital inputs were derived for
each non-traded activity. As our best estimate of the marginal productivity
of capital at market prices we employ 10 percent (net of tax). Ideally, it
would have been desirable to estimate the annual input of working capital for
firms in addition to the input of fixed capital. Unfortunately, information
on the level of working capital existed only for distributive trades; thus
working capital was excluded from all cost estimates except those for
wholesaling and retailing.
Information existed on the size and composition of the capital stock
for 15 of the 40 non-traded activities analyzed. The estimated values,
however, referred to years ranging from 1976 through 1980. To deal with these
estimates in the simplest manner possible it was assumed that the percentage
composition of capital costs in total costs in the reference year was the same
as in 1979-80. For the remaining industries the only information on capital
costs was an estimate of value added minus wages and salaries. This figure
presumably includes interest, depreciation, profits, and direct taxes. In
- 56 -
order to provide an approximate distribution of capital costs among the three
categories of domestic capital inputs an estimated breakdown of aggregate
figures was made on the basis of the average composition of assets for firms
in Egypt. The average composition of assets for firms was determined from
information on the composition of investment contained in the 1978 plan (World
Bank (1981b)). The percentage of each asset category in the total was
multiplied by its annuity factor to provide an estimate of the percentage
composition of capital costs in gross profits.
Miscellaneous-Costs and Extra Capitalists' Incomes
These two input categories are elements in the cost structure of a
limited number of non-traded activities. If extra incomes accrue to owners of
activities supplying inputs into a project under analysis, these extra incomes
should be counted as indirect costs of employing the non-traded good. While
it is not difficult to estimate the value of direct extra capitalists' incomes
resulting from:.an investment, the limited data available on supplying
industries made the estimation of extra capitalists' incomes in non-traded
activities supply inputs very uncertain. In general, therefore, it was
assumed that extra incomes did not accrue to the private owners of non-traded
activities. Any discrepancy between the estimated long run average costs of
production (including a normal return to capital) and the value of output was
assumed to consist of miscellaneous costs which were not included in the
initial cost estimate. The implication of the procedure adopted is that the
net of tax return available to investors in Egypt is close to a real rate of
10 percent. We shall argue in Chapter 3 that this assumption is quite
reasonable.
- 57 -
There were several industries in which our estimates were
particularly uncertain or in which special assumptions which departed from the
general scheme outlined above were required. These activities are discussed
in Appendix 2-1.
2.4 Household Income and Expenditure: Conversion Factors for Consumption and
Savings
Consumption Conversion Factors
Before we can proceed to estimate the accounting prices of factors of
production it is necessary to construct accounting ratios for marginal
increases in consumption by various income categories. These consumption
conversion factors will be used in Chapters 3 and 4 when we attempt to
estimate the cost in terms of the unit of account of policies which increase
the level of consumption by various households at market prices. Since
consumer expenditure represents an important commitment of resources, we
devote considerable attention to establishing the value of a unit increase in
consumption at accounting prices. The value of a unit increase in consumption
at market prices in terms of the numeraire depends upon the commodity
composition of marginal consumer expenditure. Only one reasonably reliable
source of data on consumer expenditure exists for rural and urban households
in Egypt, the Household Expenditure Survey 1974-75. The expenditure survey
covers 3000 households in rural and urban areas and represents the only
comprehensive source of expenditure data by income class for Egypt in the
1970's.
In principle, our estimates should apply to marginal increases in
consumption and therefore require the use of marginal expenditure weights.
Lacking estimates of expenditure elasticities, we have been compelled to work
with the average weights from the expenditure survey. This introduces one
important potential source of bias on our estimates of consumption conversion
- 58 -
ratios. Incomes have grown rapidly in Egypt since 1975, and it is reasonable
to expect that, particularly in urban areas, the expenditure weights for
agricultural commodities and subsidized basic consumer goods may have declined
somewhat from the 1974/75 levels. Because of the high accounting ratios
applicable to subsidized urban consumer goods, this may cause our estimates of
consumption conversion ratios to overstate the accounting cost of average
urban consumption somewhat.
The categories for which consumption conversion ratios were
calculated cover average rural and urban consumption plus high and low income
conversion ratios for urban consumers. The lower limit for high income
households was taken as LE 1000 per annum in 1974/75, which roughly defines
the upper decile of all urban households by income. The upper limit for low
income households was taken as LE 200 per annum which defines roughly the
bottom three deciles of the urban household distribution. In our general
analysis we shall use the conversion ratios applicable to average
consumption. It is useful, however, both for the appraisal of specific
projects and policies and for the insights which can be gained into the
existing tax and subsidy structure to supplement these with the conversion
factors for specific groups of households.
The expenditure distributions and the resulting estimates of
conversion ratios are presented in Tables 2-3 and 2-4. The most striking
results are the differences between rural and urban conversion ratios and the
magnitude of the conversion ratio for average urban consumption. For each LE
100 increase in average urban consumption the commitment of foreign exchange
convertable at the official rate is LE 111. An increase in average rural
consumption on the other hand commits the economy to only LE 107 for each LE
100 in expenditure at market prices. The effects of the current policy of
heavily subsidizing urban consumption are readily apparent in these numbers.
- 59 -
Evaluating Household Savings
The conversion factor for savings of wage recipients will depend on
the form in which increases in wealth are held. Because little evidence
exists on the asset choice of households in Egypt, a simple model for
determining the social cost of several forms of assets is developed, and
plausible values for the parameters are used to provide an approximate.
estimate of the accounting ratio for savings. 1/
If the savings of an average worker's household increase by one LE
(i.e., if household income increases by the reciprocal of the marginal
propensity to save), the household is assumed to hold its increased wealth in
one of three forms, as an increase in the level of cash balances, as a deposit
in the banking system, or as an investment in housing, agriculture or casual
sector activities (e.g. small scale enterprise or petty trading). 2/
1/ The 1974/75 expenditure survey contains no information on householdsavings behavior or on the asset composition of households.
2/ This simplified description of asset choice may not be altogetherincorrect for Egyptian workers' households. Land is not among assetsacquired by workers. Direct investment in industrial enterprises is alsonot a feasible alternative for most workers. Deposits in the bankingsystem, especially in the government-owned savings and commercial banks,are a form in which workers' households choose to hold their savings, andanecdotal evidence suggests there is substantial investment in agricultureby urban workers and by migrant laborers via remittances. Casual sectoractivities remain an important source of employment (although perhaps notof income) for other members of fully employed workers' households, and itseems worthwhile to include the alternative despite the absence of data.
- 60 -
Table 2-3: Accounting Ratios for Urban Consumption
Expenditure Weights 1974/75Expenditure Category Average Low Income High Income Acc. Ratio
- 200) ( 1000)
Grain and Carbohydrate .098 .182 .069 1.76
Dry Beans .015 .026 .009 1.30
Vegetables .044 .063 .031 0.92
Fruits .036 .030 .037 0.66
Meat .101 .099 .103 1.49
Fish .022 .022 .019 0.97
Milk and Eggs .050 .042 .052 0.97
Oils and Fats .041 .050 .031 1.37
Sugar and Sweets .029 .039 .023 0.84
Misc. Food .036 .059 .027 0.54
Tea and Coffee .016 .024 .012 0.54
Other Beverages .022 .007 .028 0.73
Textiles and Clothing .132 .063 .143 1.02
Housing, Fuel and Light .129 .166 .110 1.65
Furniture .036 .025 .047 0.97
Medical Care and Medicine .018 .010 .022 0.88
Transport and Communications .038 .010 .072 1.27
Education .020 .002 .034 0.86
Culture and Entertainment .013 .001 .022 0.86
Other .103 .079 .110 0.50
Consumption Conversion Factor 1.118 1.219 1.098
- 61 -
Table 2-4: Accounting Ratio for Rural Consumption
Expenditure Category Expenditure Weight Accounting Ratio
Grains and Carbohydrate .187 1.350
Dry Beans .023 .980
Vegetables .044 .920
Fruits .029 .660
Meat .111 1.163
Fish .018 .970Milk and Eggs .043 .970
Oils and Fats .058 1.300
Sugar and Sweets .038 .536
Misc. Food .022 .536
Tea and Coffee .024 .540Other Beverages .010 .730Textiles and Clothing .124 1.020
Housing Fuel Light .098 1.626Furniture .034 .970
Health and Hygiene .010 .880Transport and Communications .030 1.270
Education .010 .860
Culture and Entertainment .004 .860
Other .083 .500
Consumption Conversion Factor 1.070
- 62 -
The average saver is assumed to be infinitely lived and does not dissave.
Thus, his additional savings represent a permanent increase in the stock of
wealth and are not drawn down at a later point in life-cycle of the household
to finance consumption. 1/ The social cost of each type of asset is
separately estimated, and the social cost of savings in general may then be
found by weighting the social cost of each asset by its proportion of the
typical savers portfolio.
The social cost of an increase in cash holdings is most easily
estimated. Given the assumptions, the increase in cash balances held by the
household is permanent. Because the saver is not expected to draw down these
cash holdings at any future date to finance further consumption or the
purchase of alternative forms of assets, the social cost is zero.
Assets held in the form of investments in the banking system in
agriculture or in the casual sector represent an addition to the stock of
capital. If the unit of savings is invested in capital goods with a market
1/ The assumption of an infinitely lived saver, although not essential forthe analysis, greatly simplifies the algebra. It is consistent with alife-cycle savings hypothesis in which the amount of household incomeallocated to consumption over the life cycle of each cohort is equal tothe resources accruing to the household over its lifetime. In such amodel, households in each cohort first save and then dissave to maintain apreferred level of consumption. It has been shown that in the presence ofsteady population growth and/or growth of income, the saving andwealth-income ratio will remain constant for all cohorts taken together.Thus, the infinitely lived saver represents the average over all cohortsin the population, and if the conditions of steady growth of population orincome are approximately fulfilled, he will not dissave and his savingsand wealth-income ratios will remain constant, as hypothesized. SeeF. Modigliani, "The Life Cycle Hypothesis of Saving, The Demand for Wealthand the Supply of Capital", Social Research, 33, No.2, pp. 160-217.
- 63 -
value of K , the accounting cost of the unit of investment is equal to
fkK , where fk is the accounting ratio applicable to capital goods of
the type purchased. The investment generates an annual stream of income at
accounting prices determined by the social rate of return on the investment
activity. If the investment is in the public sector, as for example, when the
increase in savings is held in the banking system, the social return is
assumed equal to the accounting rate of interest. 1/ Hence, the annual social
income from the investment in period t is RfkKt at accounting prices when
R is the accounting rate of interest. If the assets are privately held they
are assumed to earn a real rate of return of r per cent per annum at market
prices. The annual stream of social income produced by the private casual
sector investment may be found by multiplying the annual returns at market
prices by an accounting factor for gross profits, fp. The accounting ratio
for gross profits is the ratio of value added at world prices less wages
evaluated at the shadow wage rate to value added less wages at market prices.
The act of saving and investment by the household entitles it to an
annual income equal to the return on the asset at market prices. Deposits
held in the banking system are assumed to earn a real rate of return of i per
cent per annum; the annual income from the investment is, therefore, iKt.
Investments in casual sector activities earn an annual return of rKt. The
household is assumed to consume a portion of the increase in income and to
1/ This is clearly true for government-owned banks which hold the bulk of
workers' savings. Privately-owned banks may also be assumed to lend to
the government from marginal increase in funds.
- 64 -
save the remainder in the same ratio for other forms of income. 1/ Thus, in
any year the distribution of consumption and savings out of interest income
from the initial unit of savings is given by:
iK t(c + s) c + s = 1 (3)
Savings out of interest income are assumed to be reinvested in the banking
system. If the rates of interest and the marginal propensity to save are
assumed constant over time, wealth and government investment out of the
initial unit of savings grow at a rate:
K = K (I + si)t (4)t 0
The social cost of the consumption and investment out of interest in any year
is given by:
ct = iK t(cfc+ sf k. (5)
where fc is the conversion ratio applicable to the consumption of the
household. If savings are invested in agriculture or casual sector activities
consumption and savings out of the return to the investment may be found in a
similar manner. Adding the simplifying assumption that all savings are plowed
back into the same activity, the social cost of consumption and investment in
year t arising out of the return to the initial investment is given by:
Ct = rK t(f cc + f ks) (6)
The social cost of unit of savings held as bank deposits in year 0 is
given by the accounting cost of the investment in year zero less the present
discounted value of the returns to the investment plus the present value of
1/ Direct taxes are ignored because the savers whose accounting ratio we aretrying to estimate escape the direct tax net. Because the accountingratio for taxes is zero, it would be possible to include tax payments inthe derivation of the accounting ratio for savings of workers withoutdifficulty.
- 65 -
the social cost of the additional consumption and investment out of interest
income:
f = K f - E (Rf K - C )/(l + R) (7)d O k t k t t
The social cost of a unit of savings held as assets in 'casual sector
investments is similarly given by:
fd KOfk i (rf Kt -C)(l+R)t (8)
Setting Ko = 1 and substituting from above, the accounting ratio for savings
held on bank deposits may be expressed as:
fd fk (Rf- cif - sif )/(R - si) (9)d k k c k
and for other investments as:
fd k- (rf p- crf srfk )/(R - sr). (10)
Empirical values for savings held as bank deposits and in casual
sector investments are presented in Table 2-5. In column 1 a unit of savings
is assumed to be held as deposits in the banking system. Sources of the
estimated parameter values are given in the notes accompanying the table. The
accounting ratio for bank deposits is quite robust with reference to
reasonable changes in the parameter values. Stipulating that all interest
income is consumed raises the accounting ratio only slightly to .160.
Allowing the accounting rate of interest to increase to 10 percent results in
a fall in the accounting ratio to .102.
Estimated parameter values for only three casual sector activities
are applied to the model. The activities considered are petty trading,
housing and small scale enterprise which are probably the most important uses
of investible funds in the casual sector.
- 66
Table 2-5: Accounting Ratios for Savings of Households
Notes: The accounting ratio for fk is based on the accounting ratiofor the average composition of capital stock. The real rate ofreturn on bank deposits is assumed to be 1.0 percent. Depositrates vary from 9-15 percent in nominal terms. With expectedinflation between 12 and 15 percent real rates of interest mayrange from -6 to 3.0 percent. Real rates of return to casualsector investments discussed in text.
Table 2-6: Alternative Values of the Accounting Ratiofor Household Savings
Case Case Case Case AccountingType of Asset 1 2 3 4 Ratio
1/ Compare for example the rates of 3.0 for Pakistan (Squire et al(1979)) and 4.5 for Turkey (Mashayekhi (1980)). This is a medium termparameter appropriate to the period 1983-88. The CRI is obviouslytime dependent and will vary with the growth path of per capitaconsumption. For a fuller discussion on the time path of the CRI and
estimate of its long term behavior see Dervis, Martin and vanWijnbergen (1982).
- 115 -
3.3 The Accounting Rate of Interest
The accounting rate of interest (ARI) simply measures the rate of
fall in the social value of the numeraire, uncommitted foreign exchange held
by the public sector. It provides the principal intertemporal link in project
appraisal by making it possible to express net benefits (measured in terms of
the unit of account) which occur at different times in comeasurable units.
Ideally, projects should be undertaken until the marginal rate of return on
public sector investments when they are evaluated at social accounting prices
equals the ARI. If the ARI is set too low, too many projects will be
approved; conversely, if it is set too high, there will be underinvestment.
At a full optimum with no constraints on public sector saving and investment
behavior the level of investment will be pushed until the ARI is equal to the
CR1.
In principle, the relationship between the ARI, the CRI and the
marginal productivity of capital is reasonably straightforward. The CR1
measures the rate of decline in the social valuation of average per capita
consumption. The parameter v estimated in section 3.6 measures the social
value of the numeraire relative to average per capita consumption. Hence
- v/v = ARI - CRI (5)
where v/v = (dv/dt)/v. Since the marginal product of capital, q, is a major
determinant of v, under rather stringent assumptions we can use information on
the rate of return to public sector investments at accounting prices and our
estimate of the CRI to arrive at an estimate of the ARI.
If we assume that the marginal allocation of public expenditure
between consumption and investment remains constant, the rate of fall in the
value of public sector income is given by the rate of change in social
- 116 -
valuation of public sector consumption (vc) and investment (v. ):
v/v =.s v. /v. + (l-s) v /v (6)inv inv c c
where s is the share of investment expenditure in total public expenditure.
Further, we shall assume that the Government maintains the value of public
consumption relative to the value of private per capita consumption at the
existing levels discussed in Section 3.6(a); thus Ic/vc = 0. The social
valuation of public sector investment vi falls at a rate equal to the
difference between the rate at which current public investment can be
transformed into future public income and the consumption rate of interest.
Thus: ARI = s (R - CRI) + CRI (7)
where R is the internal rate of return on public sector investment which
depends on q and the length of life of capital investment, T. A set of values
for the ARI is presented in Table 3-6. Three assumed lifetimes for investment
are considered, T = 15, T = 20 and T = , which is tantamount to an
assumption of steady state growth and reinvestment in the public sector.
Table 3-6 reveals the sensitivity of the ARI to the assumed
elasticity of the marginal valuation of income and the marginal productivity
of capital. Elsewhere in this chapter we presented our arguments for the case
that in Egypt .08 q .12 and n = 1.0. This suggests a range for the ARI of
4.7 to 7.0 percent depending on the assumed internal rate of return on public
sector investment.
- 117 -
Table 3-6: Alternative Estimates of the Accounting
Rate of Interest
Accounting Rate of InterestR N 0.5 N = 1.0 N = 2.0
-10 T = 15 5.56 3.64 5.52 9.11T = 20 7.75 4.19 6.06 9.81T = X 10.00 4.75 6.63 10.38
-12 T = 15 8.44 4.36 6.24 9.99T = 20 10.32 4.83 6.71 10.46T = X 12.00 5.25 7.13 10.88
A lower bound for the ARI is surely set by the marginal lending rates
in dollar terms discussed in Section 3.1(c). There we argued that the real
rate of interest on dollar denominated assets was in the range of 3-5
percent. The upper end of this range is consistent with our estimate of the
ARI derived above, but there is some danger that if we select an ARI as low as
say 5.0 percent too many investments projects will be approved. On the other
hand, if we set the ARI at 10 percent this implies a rate of change in v of
4.5 percent per annum which is extremely rapid. 1/
Our evidence is quite consistent in suggesting that 4.0< ARI < 8.0
and we shall therefore take as our best guess a value for the accounting rate
of interst of 6.0 percent. For projects which are very sensitive to the
assumed discount rate, alternative values of 4 and 8 percent should be checked.
1/ An ARI of 10 percent would also imply a rate of increase in the critical
consumption level of 10 percent per year which is very high given past
performance on the growth of per capita consumption. On this point seeScott, Mac Arthur and Newbery (1976) p 43ff.
- 118 -
3.4 Relative Weights for Per Capita Consumption
The methodology of estimating income distribution weights is now
quite well established; (see for example Squire and van der Tak (1975), Scott,
McArthur and Newbery (1976) or Squire, Little and Durdag (1979)). 1/
In order to estimate the social benefit which accrues when a
household receives an increase in disposable income, a number of assumptions
concerning the Egyptian Government's attitude toward increases in the incomes
of private citizens are required. One of the simplest, and most conventional,
of assumptions to make is that planners assign progressively lower marginal
valuations to income as household incomes rise. Most social cost-benefit
studies have employed a marginal valuation function of the form:
U'(Y) = k Yn (8)
where the single parameter n, the elasticity of the marginal valuation of
income, summarizes the degree of egalatarian bias of the Government. By
arbitrarily setting U'(Y) = 1, where Y is the mean level of income, it is
possible to derive the value of the scaling factor k and the relative weights
to be assigned to marginal changes in income at varying levels;
d = U'(Y) = Yn (9)y
Thus, given estimates of the mean level of income Y and the elasticity of the
marginal valuation of income, n, it is possible to derive a consistent set of
welfare weights reflecting the social valuation of marginal changes in private
incomes.
1/ Because many of these concepts are more fully developed in Squire and vander Tak the reader may refer to their work for additional exposition ifdesired.
- 119 -
Because data on incomes by household are lacking in most developing
countries, it has become conventional to work with expenditure surveys and to
use consumption expenditure as a proxy for income. This is the approach which
we have adopted for Egypt as well. Tables 3-7 and 3-8 present the estimated
values of the relative welfare weights assigned to marginal increases in
household incomes by income class for both rural and urban households, based
on the 1974/75 expenditure survey. Interpretation of the tables is
straightforward. Because household size varies with income, we have chosen
household consumption per capita as the appropriate indicator of income
levels. 1/ The positive association of household size with income means that
the household income distribution tends to overstate the inequality of income
and hence the dispersion of the income weights relative to the per capita
distribution. Three possible values of the elasticity of the marginal
valuation of income, n, are listed in the tables. Given that Egypt employs a
progressive income tax system it is doubtful that the appropriate value for n
is less than or equal to 0; for this would imply that the Government places
equal weights on marginal increases in the incomes of all individuals or in
the case of a negative value of n that it values marginal increases in the
incomes of the rich more highly than those accruing to the poor.
The relation between n and the weights assigned to marginal increases
in income is quite clear. With n = .50 the marginal benefit of increasing the
consumption of a household in the lowest income class by one Pound is slightly
1/ It would have been more preferable to work with income per adultequivalent, which would have permitted correction for householdcomposition as well as size, but this is not possible on the basis of the1974/75 survey.
- 120 -
more than three times the marginal benefit of increasing the consumption of a
household in the highest income class. With n = 2 on the other hand, the
welfare weight assigned to the lowest income category is more than 100 times
the weight assigned to the highest. These values probably bound the
appropriate parameter estimate for n in Egypt, and we shall consider below
evidence on its most likely value.
The relative weights summarized in Tables 3-7 and 3-8 apply to gains
(or losses) in income occuring at urban prices. If rural and urban prices
differ, the command over real consumption represented by LE I at urban prices
may differ greatly from that represented by LE 1 at rural prices. Since
Table 3-7: Relative Weights for Marginal Changesin Urban Incomes at Different Income Levels
Income Class Percentage Total Per Capita Expenditure WeightLE Per Annum Households Consumption n = 0.5 n = 1.0 n = 2.0
N4ote that all values are measured in terms of urban prices. Thus the
expenditure weights measure the relative value of a transfer of LE 0.91 at
rural prices or LE 1.0 at urban prices.
- 122 -
Suppose that the ratio of rural consumer prices to urban prices is p;
then the rural consumption level C is equivalent in real terms to an urban
consumption level of C = Cr/p. The appropriate income weight to be
assigned is therefore d = Ur~~
It is also important to recall that the value of any transfer in
nominal income between the rural and urban sectors of the economy must be
expressed in the same set of constant prices. Consequently, a unit marginal
increase in the income of a rural household has the same value as an increase
of l/p accruing to an urban household. The welfare value of a unit increase
in the income of a rural household with consumption level C expressed in
terms of urban purchasing power is therefore:
i n (10)
Table 3-8 expresses the weights assigned to marginal additions to rural
incomes measured at urban prices. We shall throughout this study measure all
incomes and gains in income at urban prices. Rural-urban price differentials
are discussed in Appendix 3-1.
3.5 A Possible Value for n
Although it is reasonable to assert that the Government of Fgypt may
attach lower value to marginal additions to income for citizens in high income
brackets, it is difficult to establish the precise value of the elasticity of
the marginal valuation of income (n). Certainly, many public policy
statements could lead to the conclusion that the Egyptian Government has a
strongly egalatarian bias. Such assertions are buttressed to an extent by the
system of urban consumption subsidies and by the progressive nature of the
system of direct taxes. The picture is obscured, however, when one considers
the overall incidence of the tax system. Indirect taxes via the system of
- 123 -
agricultural price differentials fall disproportionately on rural households
in the low and middle income range and the overall incidence of rural taxation
crucially depends on the access of rural households to subsidized farm inputs
and consumer goods. (See el Edil (1980) and Korayem (1980)).
Other writers on welfare weights have suggested that a plausible
range for the elasticity of the marginal valuation of income is between 0 and
5. The only "objective" estimate of this elasticity available for Egypt can
be derived from the structure of the income tax system, based on the
assumption that the tax tables reflect the principle of equal marginal
sacrifice. Marginal sacrifice, the welfare loss to an individual due to taxes
collected from an incremental LE 1.00 of income, may be expressed as:
MS. = u(Y.) t. (11)1 1 1
where MS. is the marginal sacrifice of household with income Y., U'(Y.)31 1 1
is the marginal valuation of income for a household with income Y. and t.
is the applicable marginal tax rate. If the Government follows a principal of
equal marginal sacrifice, the MSi are equal for all income intervals i.
Thus, taking the lowest income class subject to direct taxation as a base and
specifying a marginal valuation function of the constant elasticity form in
(8) we may write:
Ms~ kY -n t MS ~-nt
Msb kY -n tb (12)b
Simplifying, taking logarithms, and rearranging terms:
ln(ti/tb) = -n ln(Yb/Yi) (13)
Equation (13) was fit to the Egyptian 1979 tax table by ordinary least
squares. Income was taken as the midpoint of the range applicable to each
- 124 -
(a) The Social Value of Consumption Subsidies
Egypt has a long-standing commitment to a system of consumer
subsidies which is one of the most extensive in the non-centrally planned
economies. The subsidy bill ranged from 8 percent of CDP in 1977 to 11
percent in 1979 and accounted for approximately one third of central
government current expenditure in both years. Major consumption subsidies
exist for a wide range of staple food items including, wheat, corn, bread,
beans, oils, meat and sugar. Subsidizing these commodities represents an
effort by the Government of Egypt to insulate urban consumers from the
effects of rising import prices. In choosing the extent of the subsidy the
Government has made an implicit judgement regarding the value of public
income relative to private sector consumption. Thus, from a detailed
examination of the incidence of the subsidy it is possible to estimate the
value of public income committed to the subsidy program. Using our
estimates of the relative welfare weights for marginal additions to
consumption (d) derived in Section 3.4, it is possible to estimate the
marginal benefit, the increase in consumer welfare relative to the average
level of urban per capita income, of an expenditure of LE 1.00 on
additional consumption subsidies. The cost of the subsidy -- that is the
reduction in uncommitted public sector income denominated in terms of
foreign exchange -- is given by the consumption conversion factor f .
Thus, the welfare valuation of a marginal unit of public income committed
to consumption subsidies is given by the ratio of the marginal welfare
benefit to the marginal resource cost:
Vj i-l wiji]fc (14)
where v. is the marginal valuation of public income committed to
- 125 -
specified marginal tax rate. Marginal tax rates range from 4 percent for an
annual income of LE 325 per year to 92 percent at an annual income of
LE 8500. Results of the regression are reported in Table 3-9. Rather than
supress the intercept, we have estimated a standard OLS regression which
permits us to test for the presence of an intercept significantly different
from zero.
Table 3-9: An Estimate of n from the Egyptian Tax Tables
Coefficient Standard Error
Constant -0.014 0.095
ln (yb/yi) -0.957 0.044 R2 = .978 Obs = 13
The results yield a statistically significant estimate of n .96, an
intercept not significantly different from zero, and show a remarkably good
fit. This is undoubtedly an overestimate of n, since we have made no
attempt to allow for exemptions and other exclusions of income which might
lower the marginal tax rates in upper income brackets. It does suggest,
however, that the upper bound for the elasticity of the marginal valuation
of income should be set at 1. We have, accordingly, taken 1 as our
principal estimate. For comparative purposes in some calculations we have
also employed values of n = 0.5 and n = 2.0.
3.6 The Value of Public Sector Income (v)
Our unit of account is public sector income, denominated in terms
of foreign exchange. Because we wish to express the value of changes in
private incomes relative to this unit of account, we require an estimate of
the value of public sector income (the numeraire) relative to marginal
addditions to private consumption at the average level of urban per capita
- 126 -
consumption (income). With this parameter (v) it is then possible to
transform the relative income weights derived in section 3.4 into a set of
absolute income weights: w = d/v.
Because at the margin public sector income may be used for a wide
range of activities -- investment, education, defense, consumer subsidies,
general administration etc. -- the appropriate value for v is the weighted
average of the relative values of public expenditure on these items to
average per capita consumption. As a practical matter it is difficult to
establish relative valuations for such activities as education, general
adminstration or defense. In principle, however, a rational government
will allocate expenditure to equalize the marginal social benefit of all
areas of expenditure. In these circumstances, an estimate of one component
of v is sufficient to provide information on the relative value of the
numeraire in any other activity. It is doubtful that Egypt has explicitly
followed such a course of public expenditure analysis, particularly with
regard to defense, education, and general administration. Lacking
sufficient information to provide independent estimates of the value of
expenditure in these areas, however, we shall focus on three areas of
public expenditure where there is sufficient information to estimate the
relative value of public sector income; consumption subsidies, public
sector employment policies, and public sector industrial investment. Since
the Government of Egypt undoubtedly enjoys greater budgetary discretion
over these three items than over defense and general administration, they
may comprise the main component of marginal public expenditure, and hence
provide an appropriate estimate of v.
- 127 -
subsidization of commodity j, w.. is the expenditure weight on commodity
nj of expenditure group i, E w - 1 and d. is the welfare weight for
i ij I
expenditure group i relative to that for average per capita consumption in
the urban sector.
The calculations of the marginal benefits for subsidies of grains,
oils, meat, beans, and sugar are set out in Table 3-10. In columns (1) and
(2) we present the distribution of urban households by income class.
Columns (3)-(5) present the relative income weights by class of household
for three values of the elasticity of the marginal valuation of income, and
columns (6)-(10) provide the consumption weights for each income class for
each of the five subsidized commodities. The marginal welfare benefit is
found by multiplying the expenditure weight by the relevant welfare weight
and summing over all expenditure categories. The results show a value of
the transfer relative to average per capita income ranging from .967 for
meat subsidies (n = .5) to 1.61 for beans (n = 2). At our preferred value
of n = 1 the value of the transfer lies in the interval 1.00 - 1.17
depending on the product category.
The relative rankings of the welfare value of the transfer are
similar under any assumption regarding the elasticity of the marginal
valuation of income. The greatest welfare benefits accrue to commodities
which have the highest expenditure weights for low income households, beans
and grains, and the lowest benefits accrue to high income consumption goods
such as meat and sugar. Indeed, in the case of meat subsidies the
Government could achieve approximately the same welfare benefit by
transferring income directly to individuals at random.
Table 3-10: The Social Value of Urban Consumption Subsidies
Income Class Percentage of Relative Expenditure Weight Proportion of Total Consumption of Subsidized ItemLE Per Annum Urban Households N = 0.5 N = 1.0 N = 2.0 Grains Beans Oils Meat Sugar
Total 7/ 7,832.0 11,131.6 2.2 100.0 100.0 3,299.6 100.0
of which (288.9)1/ 1,779.0 (12.9)3/ (3.7)1/ 16.0 (1,490.1)3/ (45.2)Government (932.9)2/ (6.3)4/ (846.1)4/ (25.6)
1/ 1961; comparability with the 1976 figures is uncertain.
2/ 1965/66; comparable.
3/ 1961-76.
4/ 1965/66-75.
5/ Including unemployed.
6/ Not including military conscripts and public enterprise employment both of whichare reported in industries 1 to 10.
7/ Including military conscripts.
Sources: Population censuses, adjusted for comparability; see A. Nassef, 1980. ForGovernment, 1965/66-78 see Central Agency for Organization and Administration,"Development of Employment in Government Administration, 1965/66-1978",Information Centre, Data Series No.11, September 1979, p.15.
liible 4-2: Ib,or Force in Egypt by lnda,itry, 171-79
Number of Persons Annual Shares of Shares of increase Shares of Increasethousands Rates of total, 2 1960-76 in Employment
Industry Growth, Outside Agriculture
1971 1979 Compound 1971 1979 Number,% Thousands x x
Sources: Col. 1: CAPMAS, Private and Public 1975, public only weekly wages. Daily wages obtained byassuming six working days.
Col. 2-5: K. Fodah, Productive Efficiency in the Contracting Sector, quoted by ... . Daily wages.Based on a survey of contractors.
Col. 7: N. Chouri, R. Eckaus, A. Mohei-Eldin, "Migration and Employment in tbe ConstructionSection ... ", Ch. 8, MIT, 1978, p.1 09.
Col. 8: Ministry of Agriculture.
- 155-
and government, and a major widening of the differential between agriculture
and construction, reflecting the influence of emigration on the wage
structure. The average differential between agricultural employment and
manufacturing remains large, approximately 100 percent. We lack reliable
comparisons for unskilled labor, however, a problem which greatly complicates
our estimate of the shadow wage.
Table 4-5: Sectoral Wages, 1966 and 1975
Average Daily Wages1966 1975
PT Index PT Index
Agriculture, men 25 100 46.5 100Construction, excavators, men 25 100 100.0 215Manufacturing industry
Public 70.5 282 105.1 226Private 46.5 186 85.9 185
Government 103.3 413 127.9 275
Source: ILO(1980). It is assumed that employees in the manufacturingindustry work 6 days per week and government employees 300days per year.
4.2 A General Formula for the Shadow Wage
Consider a project which employs an additional worker. In principle,
the worker could be drawn from anywhere in the economy, and the social cost of
employing him in sector i is given by:
SWR. = z p.(M. + C. - B.) (1)1 J J J J J
where
p. = the probability that the worker will be drawn from sector j
into sector i;
- 156 -
M. = the value at accounting prices of the output foregone by
withdrawing a worker from his previous occupation i into
occupation j;
C = the social cost in terms of the numeraire of providing the
additional,resources required to meet any extra consumption,
and savings of the worker and his household arising from the
excess of wages in sector (i) over wages in his previous
employment in sector (j), from any changes in the incomes of
other household members, and/or from any increase in the
disposible incomes of workers in sector (j) as a result of
bidding up of wages in sector (j);
Bj = the social benefit evaluated at accounting prices of the
increase in disposible incomes of workers transferring from
sector j to sector i, from a change in the income of other
members of the worker's household, and from any transfer of
income to workers remaining in sector j as a result of the
bidding up of wages.
The values which are assigned to each of the elements of the shadow
wage and the way in which the probabilities pj are estimated depend upon the
character of the shadow prices to be estimated and upon the assumed nature of
the labor market.
It has become usual in Bank practice to distinguish between shadow
prices calculated on the so-called "efficiency" basis.and social accounting
prices. The distinction between them is that efficiency prices measure the
traditional opportunity cost of using a commodity or primary input while
social accounting,prices incorporate distributional considerations, either
- 157 -
among contemporaries or over generations. Substantial time and ingenuity has
been devoted to the debate over the relative merits of the two systems,
largely with the result that the issues have become less rather than more
clearly defined. It should be apparent that the whole approach of this paper
has been to argue that project appraisals - particularly in an economy like
Egypt with a history of policy interventions affecting income distribution -
should be based on a system of social prices which reflect the relative
priorities assigned to various policy objectives. Bank practice, however has
traditionally emphasized the use of efficiency prices. For most commodities
the differences between social and efficiency prices is negligible, but with
respect to the shadow wage the distinction is very important. Thus, we shall
examine the components of the shadow wage expressed in equation (1) which make
up efficiency and social prices.
Traditional efficiency prices have been based on estimates of the
value of the marginal product of labor in its alternative employment expressed
in terms of the unit of account. Thus the traditional efficiency wage is:
SWR. = p.M = .p.( f.m.) (2)i i ii iJ J J J
where f. is the accounting ratio assigned to the output of sector j and m.
is an estimate of the marginal value product of labor in sector j at market
prices. Note that this formulation introduces a very powerful assumption:
--p.( C. -B.) = 0 (3)
that is that the net resource costs of providing additional employment are
confined to the fall in national income evaluated at border prices resulting
from a change in the existing allocation of labor. This is appropriate if
C. = B. for all j, which implies that the social cost of the resources
required to finance the increased consumption of each class of households is
- 158 -
precisely offset by the social benefit of the increased income. This is not .a
weak distributional assumption; rather it is a very strong one. It implies
that at the margin public sector income is no more or less valuable than the
income of any potential private recipient. Such an assumption is inconsistent
with the whole tenor of our evaluations of tax and subsidy policy in Egypt and
with the policy debate surrounding the need to increase public sector savings
and investment, as well as with the received wisdom of the theory of planning
and optimal growth.
Even if we were to assume that wages were the same in all sectors of
the economy and that the creation of a new public sector job did not result in
any change in household incomes, if the new job resulted in the migration of a
worker from rural to urban areas (or vice versa) the combination of different
expenditure patterns and differential taxes and subsidies mean that the
resource cost per unit of rural and urban expenditure differ substantially.
It seems inappropriate to consider the marginal product of labor as a
legitimate efficiency cost but to ignore the resource costs associated with a
change in the pattern-of household expenditure. In Egypt, for example, the
resource cost associated with creation of an urban job is substantially
greater than that associated with the creation of a rural job. Shadow pricing
should reflect this phenomenon and guide investment accordingly.
To this end we also consider what have come to be called "extended
efficiency prices". The extended efficiency shadow wage is written as:
SWR. = £p. (M + C. -i J J :i J U
= Ep. (f.m. + f .Y.-f .Y. - b(Y. - Y.) (4)j jj cii C] j j
where fcj and fci are the conversion factors applicable to consumption in
sectors j and i respectively, Yi and Y. are the household incomes in
- 159 -
present (sector i) and former-(sector j) employment and b is the single income
weight applicable to the income of all households. This parameter is equal to
the inverse of the parameter v, the value of uncommitted public income
relative to urban average per capita consumption. (Note that as v becomes
large, b declines and the benefit associated with the increase in consumption
tends to zero.)
It should be clear from (4) that even if Y. = Y. the extended1 J
efficiency shadow wage will only equal the efficiency shadow wage if
fci=fcj. In all other cases, the extended efficiency shadow wage will
reflect changes in the composition and accounting cost of consumption
expenditure even for equal household incomes.
Finally, we consider the social accounting wage. This version of the
shadow wage rate incorporates a full set of differential income weights by
class of household. Hence
SWR. = p.(FM. + C. - B.)1 J]I J J J
= pj( f m.+f .Y.-f .Y.-b..(Y.-Y.)) (5)i j ci. I C] j ij 1 j
where bij is the welfare weight applicable to the change in income
(Y. - Y.).
4.3 The Shadow Wage for Skilled Labor
In setting out the general formulae for the shadow wage in section
4.2, we noted that the parameter values applicable to each of the components
of the shadow wage depend upon the structure of the market for the relevant
type of labor and on the assumptions made concerning the way in which that
market adjusts to marginal increases in the demand for labor. In this section
we consider a very simple model of the market for skilled labor in Egypt which
- 160 -
nonetheless captures the essential elements of adjustment to the high levels
of external migration of skilled labor discussed in section 4.1.
Our evidence on the structure of the market for urban skilled and
semi-skilled labor leads to the conclusion that within the Egyptian economy
wages for these categories of workers are reasonably competitively determined,
and since there are no domestic restrictions on emigration it may be argued
that the labor market for skilled and semi-skilled workers is in equilibrium.
Thus, the marginal migrant should be indifferent between emigrating and
remaining in Egypt. There is, however, a substantial differential in the wage
rates paid to homogeneous categories of labor in Egypt and in the Gulf states.
We, therefore, require a model of the labor market which is
consistent with our assumption of competitive equilibrium and with the
persistence of an Egyptian-foreign wage differential for identical categories
of labor. Such a model of the labor market is depicted in Figure 4-1. We
assume that the supply of skilled labor is fixed at the level 00'. The
foreign demand curve for Egyptian skilled labor is given by DfDf and the
domestic demand curve by D D . The effective supply curve of Egyptian
skilled labor to foreign employment is given by S S which lies above the
marginal value product curve D D . Thus we have assumed that the supply
price of Egyptian labor to foreign demanders exceeds its marginal value
product (assumed equal to the wage) in Egypt, due for example, to such factors
as imperfections in information flows, the psychic disutilities of foreign
migration, the real resource costs of moving from one sector to another, and
the fact that migrant jobs are not tenured and involve eventual return
- 161 -
migration. 1/ Foreign demanders hire Egyptian labor equal to O'L at a wage
W . Domestic skilled labor OL receives a competitively determined wage W
and there is a foreign domestic wage differential WfWe.
5E~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~S
We.
0 L L' ,
Figure 4-1: Equilibrium in the Labor Market for Migrants
The shadow wage for skilled labor depends on the process of
adjustment of this labor market to an increase in the demand for skilled labor
in Egypt. Assume the demand for skilled labor shifts outward by LL'. The
reservation wage for Egyptian skilled labor supplied to foreign demanders
1/ Indeed, a more complete model would treat the migration decision as anintemporal optimization problem in which the marginal migrant must equatethe present discounted values of the stream of benefits and costs ofmigration job search and return to the opportunity cost of remainingemployed in Egypt. The qualitative results of such a model, under
reasonable assumptions, lead to a supply price in excess of the static
wage.
- 162 -
increases accordingly and the effective supply curve of migrants shifts to
S S . This will raise the wage rates both in the domestic economy ande e
abroad. The increased demand for labor in the domestic economy will be met by
a reduction in the volume of migrants who are obtained by the bidding up of
domestic wages as well as by the deprivation of other producers in the
domestic economy. The proportions in which workers are drawn from foreign
and domestic employment into the new project will depend on the elasticities
of foreign and domestic demand, on the initial allocation of labor between
foreign and domestic employment and on the elasticity of supply of migrant
workers. The shadow wage will consist of the cost to the economy of foregoing
L"L migrants plus the cost of withdrawing L"L' workers from other domestic
employment plus the resource cost (net of any income benefit) of the increase
in domestic consumption made possible by the bidding up of the domestic wage
w we w
Let Vf and yd be the elasticities of foreign and domestic demand
(defined so as to be positive) respectively and s be the elasticity of
supply of migrants to foreign employers. Then the proportion of workers drawn
from foreign employment is:
Pf -(6)
f Sf rff + (l-sf)c (6)
where sf is the proportion of the Egyptian skilled labor force working
abroad. The proportion drawn from domestic employment is:
(1-Pf) = (7)Sf fl + ( 1 s)C
and the increase in the wage rate is:
We dE (1-pf)We dEsflf + (1-sf)T E (1-sf)% E (8)
where W is the domestic market wage and dE/E is the percentage change in
demand for skilled labor.
- 163 -
It is now possible to consider the social cost of foregoing an
incremental migrant. When a skilled laborer is withdrawn from overseas
employment there is a reduction in Egyptian national income denominated in
terms of foreign exchange equal to the value of his annual remittances (rem).
Hence M = rem.m
There is also an increase in the committment of resources to
consumption and private savings equal to the difference between his domestic
wage and the amount of total remittances. The resource cost of meeting this
change in the household income of the migrant and his family is therefore:
Cm = W (6 kf +(l-6 sk)fc) - (rem)(6 remf s+(- rem )c (9)
where 6sk and 6 e are the marginal propensities to save out of income of
households at the level of income associated with skilled workers and out of
remittance income respectively and f and fc are the conversion factors
for savings and consumption as defined in Section 2.3.
Finally, we consider the benefit from the change in household
income. Given our assumptions of equilibrium in the labor market, there is no
net benefit from the change from foreign to domestic employment by the
marginal migrant. The change in wages is precisely offset by the changes in
conditions of employment, disutility of migration, differences in real incomes
etc. Hence we set B = 0.m
We next consider the cost of drawing skilled labor from the existing
allocation in domestic employment. If we assume that the labor is drawn from
a variety of industries, both traded and nontraded, it is probably most
appropriate to identify the value of output foregone with the market wage
multiplied by the standard conversion factor (f ). Therefore;
M =W (f ). (10)sk e sc
- 164 -
The bidding up of the domestic wage also results in a transfer of
income from the consumers of output produced by skilled workers or from their
employers to the workers employed in the domestic economy. The value at
market prices of the transfer is:
(l-sf ) WEddW dE (11)
dd
where E d is the level of domestic employment of skilled labor. The
consumption cost of this transfer per worker (dE = 1) is:
Csk = EddWe (f ct-fc) = ((S)fWe/Id)fct -fc) (12)
where f is the conversion ratio for consumption applicable to those whoct
bear the transfer. In general, we shall assume fct = fc and hence that
sk
Finally, we consider the consumption benefit of the transfer:
Bsk = L(( sf) We)/11U](bt-b k) (13)
where the parameters bt and b k are the welfare weights assigned to the
incomes of those bearing and receiving the transfer respectively.
We put together our best estimates of the various parameters of the
shadow wage for skilled labor in Table 4-6. First we consider the opportunity
cost of foregoing one additional migrant. Average remittances for migrant
workers in 1975 - 1978 (expressed in constant 1975 prices) are LE 917. The
resource cost of additional consumption is based on two important
assumptions. First, we take as the market wage of a typical migrant in Egypt
(W ) an annual value of LE 1092 (at 1975 market prices) which is the average
of the daily rates for skilled construction workers (masons, carpenters, and
plumbers) at constant 1975 prices over the period 1975-1977. We also make the
important assumption that marginal savings out of remittance income are
significantly higher than those out of domestic employment income.
Table 4-6: Shadow Wages for Skilled Labor Drawnfrom Various Sectors of Origin
Sector of Origin Social Value of Resource Cost of Social Benefit of Total Netof Skilled Labor Decline in Output Change in Consumption Change in Consumption Social Cost
Mj Cj Bj SWR
REM 9/7 sk = .25 SWR = 1206
Migrant Worker f = .60 SWR = 1.104
fc = 1.10
Mm = 917 km = .50 Cm = 289 Bm = 0
Sf = 150 SWR = 1332
Other Domestic We 1092 d = 1.000
Employment fsc= .965 bt .80 SWR = 1.220
MSk = 1054 Csk = 0 bsk = .50 Bsk= 278
- 166 -
The assumed marginal savings rate out of remittance income of 0.5 is
consistent with the assumptions of the DRM model and with recent evidence
regarding the commodity composition of own exchange imports which are more
than 50 percent producer's goods. Our assumed value of the marginal
propensity to save out of domestic employment income of 0.25 may be too high,
since as we noted in Chapter 2, we lack any empirircal basis for estimating
household savings. If the expenditure elasticity of savings is approximately
1.5 (the value employed by Taylor in World Bank (1980b), a marginal propensity
to save of .25 would be consistent with a household income in the range LE
1000 - 1399, and we therefore adopt that value. By assumption we have set
B = 0.m
The net resource cost of foregoing one migrant is therefore LE 917 +
289 = 1206, which is our estimate of the shadow wage of a marginal worker
drawn out of migration (at 1975 constant prices). The ratio of the shadow
wage to the market wage is 1.10. This accounting ratio is applicable to
contemporary market price data on skilled labor costs.
We now turn to an estimate of the resource cost of withdrawing a
skilled worker from domestic employment. To find the social value of the
decline in output we multiply our estimate of the market wage (LE 1092) by the
standard conversion factor of .965. By assumption there is no resource cost
of the change in the employment, since pre- and post change household incomes,
behavioral parameters, and accounting ratios are identical.
To estimate the extent of the income benefit (or cost) which results
from any bidding up of wages in the domestic economy we assume that skilled
workers fall in the household income category LE 1000 - 1399. The
distribution weight appropriate to recipients of the transfer is, therefore,
- 167 -
.50. Next, we need an estimate of the distribution weight for those losing
income. If wage increases are passed through to consumers, the impact will be
to transfer income from households in proportion to their share in total
consumption expenditure. The income weight appropriate to the expenditure
weighted average of urban households is .803. This is probably an
overestimate of the welfare loss of a marginal increase in skilled wages. It
may be argued that high income consumers account for a disproportionate share
of the expenditure on skill intensive commodities. Secondly, not all of the
wage increase may be passed on. That portion of the wage increase which comes
from private profits will have a welfare cost which depends on the levels of
savings and consumption by private entrepreneurs and on the social returns to
private investments. The portion of profits consumed will receive a
relatively low weight, if capitalists are among the high income groups. But
the portion committed to savings may carry a relatively high weight, if the
social returns to private sector investment are high. Finally, if the wage
increases affect the public sector as well, they may lead to a reduction in
uncommitted social income, provided that the price increase is not fully
passed on.
All of the possibilities described above are likely to occur to some
extent. Since we have little concrete information regarding which outcomes
are most likely, we have experimented with two values for bt of .80 and
.50. In the latter case, there is no benefit from the transfer; in the former
case, since the welfare loss of those paying the transfer exceeds the gain to
skilled workers and their households, the effect of the transfer is to
increase the social cost of skilled labor.
- 168 -
Since the extent to which wages are bid up is a function of the
proportion of workers drawn into employment from the domestic economy and the
elasticity of domestic demand, the final estimate of the shadow wage is
sensitive to the assumed value of these two parameters. The shadow wage rate
for skilled labor as a proportion of the market wage may be written as;
SWRk bt
We = 1.10 Pf + (1-pf) if + ] (14)
where btf is the net welfare weight assigned to the transfer, (b t-b k).
The range of plausible values for pf, the proportion of workers
drawn from foreign employment, is explored in Table 4-7. The crucial
relationship involved here is that between the elasticity of foreign demand
and the elasticity of domestic supply of migrants to foreign demanders. It
seems reasonable to assume that there is a high elasticity of foreign demand
for Egyptian migrants, but it may not be perfectly elastic. In order for the
proportion of workers drawn from-foreign employment to exceed .80 the
elasticity of foreign demand must exceed the elasticity of supply of migrants
by a ratio of 22:1. This is not an implausible value and is consistent with
Table 4-7: Estimates of the Proportions of Workers Drawnfrom Migrant Employment Pf
Elasticity of Domestic Elasticity of Foreign Demand
Supply to Foreign DemandrLs 5.0 10.0 20.0 50.0 .
0.5 .64 .78 .88 .95 1.0
1.0 .47 .64 .78 .90 1.0
1.5 .37 .54 .70 .86 1.0
2.0 .31 .47 .64 .82 1.0
Sf= .150
- 169 -
elasticities of foreign demand for Egyptian migrants in the range of
10-50. 1/ Thus we choose values of pf equal to .8 and .9.
Estimates of the shadow wage (as a percentage of the market wage) for
alternative values of the percentage of workers drawn from migrants and the
elasticity of domestic demand appear in Table 4-8. To interpret the table
note that if foreign demand for migrants is perfectly elastic pf = 1.0 and
SWR/W = 1.10 if it is zero our best estimate of SWR/W is' 1.22. The
range of values depending on the elasticity of domestic demand and the
proportion of foreign workers drawn into domestic employment is relatively
compact, lying in the interval 1.07 - 1.19. We select as our best estimate of
the ratio of the shadow wage to the market wage a value of 1.08 at efficiency
prices and extended efficiency prices and of 1.12 at social accounting prices.
Table 4-8: Sensitivity of the Accounting Ratio for Skilled Laborto Estimated Social Values of Income Transfer andProportion of Foreign Workers
Pf = .80 Pf = .90
btfr .5 t=1. 0 r<k= 1. 5 yk_= .5 1_ 1.0 -rL- I. 5
0.30 1.193 1.133 1.113 1.146 1.117 1.107
0.00 1.073 1.073 1.073 1.087 1.087 1.087
1/ Recall that the elasticity of demand for Egyptian migrants is givenby (Wkw+(l-a)Ew)/a, where > is the elasticity of demand formigrants from all points of origin, Sw is the elasticity of
supply from alternative suppliers of migrant labor, and a is theproportion of total migrants supplied by Egypt. If a isapproximately .10 and ifvw andew both equal 2, the elasticityfacing Eygpt is 38.
- 170 -
4.4 Shadow Wages for Unskilled Labor
In this section we present estimates of the shadow wage rate for
urban and rural unskilled male labor. The focus of our effort is to determine
the social cost of employing an additional unskilled worker in a public sector
project.
For the purpose of this analysis the Egyptian economy may be divided
into four sectors. The formal sector is assumed to consist of private and
public enterprises which are subject to the minimum wage. The sector roughly
encompases urban and rural enterprises employing 10 or more workers which are
engaged in "modern economic" activities. Total employment in the sector in
1976, the last year for which complete data are available, was 1.03 million.
The distribution of employees between rural and urban enterprises and among
categories of industry is estimated in Table 4-2. The casual or informal
sector consists of traditional non-agricultural activities such as crafts,
petty trading and personal services. Very little is known about employment
and incomes in such activities in Egypt.
The agricultural sector in Egypt consists almost entirely of small
and medium holders engaged in export or food crop production, and casual or
permanent agricultural workers. Employment in export production occupies
approximately one-third of the agricultural labor force and food crop
production occupies the remaining two-thirds. Hired labor is estimated
variously at between 13 and 15 percent of the agricultural labor force.
Unemployment in rural and urban areas forms the forth sector considered. Here
again data are very scarce, but the best estimate available places the rate of
unemployment in 1979 at 4.5 percent of the labor force. (ILO (1980)).
- 171 -
At this point it is useful to introduce and discuss several
assumptions which simplify the application of the general formula for the
shadow wage. These assumptions specify the mechanisms of adjustment of the
labor market to an increase in the demand for unskilled labor, and therefore
define the values of the parameters for the estimates of the social cost of
labor drawn from each source. First we shall in general assume that new
workers in the formal sector orginate in one of three other sectors,
agriculture, the casual sector, or the pool of unemployed. Labor could be
drawn from other employment in the rural or urban formal sector, but in these
cases it seems reasonable to assume that any vacancies will be filled by
workers drawn out of one of the three sectors mentioned. Provided that the
transfer of occupation within the modern sector has no effect on formal sector
wage rates, the social cost of obtaining a worker from elsewhere in the formal
sector may be estimated by finding the cost of his replacement. Thus, he may
be regarded as having been indirectly drawn from one of the three primary
sources of modern sector labor. There is one important exception to this
assumption. As we noted in section 4.1, there is substantial overstaffing of
many public sector industrial enterprises. If the creation of a new formal
sector job allows other enterprises to release redundant labor, the social
cost of that labor will be less than the wage. Thus, we also consider the
case in which a proportion of workers drawn into new formal sector employment
originate in the pool of redundant workers in the public sector.
Second, we shall assume that drawing a worker from the casual sector
into formal sector employment will result in his immediate replacement by
another workers from the agricultural sector or from unemployment. In this
case, withdrawing a worker from the casual sector is analytically equivalent
to directly drawing his replacement from agriculture or unemployment into the
- 172 -
formal sector. Given the ease of entry from agriculture into casual sector
activities and the high rate of rural to urban migration, this assumption may
not be far from accurate. If it is not wholly accurate, it involves making
two offsetting errors. To the extent that withdrawing a worker from the
casual sector results in a net decline in casual sector output -- as for
example, when he is not completely replaced either .by the transfer of another
worker from agriculture or unemployment or by increased effort on the part of
remaining workers in the sector -- the price of casual sector goods should
rise. This will result in a rise in earnings from casual sector employment, a
transfer of income from consumers to producers of casual sector output, and in
a decline in the amount of casual sector output demanded. Disregarding the
net decline in casual sector output and neglecting the resource cost imposed
by any excess of casual sector wages over wages in the sectors from which new
casual sector workers are drawn involve neglecting a cost. Failure to fully
credit the benefit from the increase in disposable incomes of workers drawn
into the sector and from the transfer of income from consumers to casual
sector workers involves neglecting a benefit. 1/
Third, in estimating the social cost of a worker drawn into formal
sector employment, the effect (if any) of his departure on wages in the sector
from which he is drawn is assumed to be sufficiently small to be ignored.
The assumption allows us to take as the shadow wage rate in the sector from
which the worker is drawn, Mi, the social value of his marginal product.
1/ When the limited data are applied to an analysis similar to that in theappendix, we find that disregarding a potential fall in net output fromthe casual sector will result in an understatement of the "true" shadow
wage of workers drawn from the informal sector of less than 5 percent.
- 173 -
In the appendix to the chapter a general formula for the shadow wage is
derived for cases in which the market wage is bid up. 1/
(a) The Social Cost of Output Foregone
The preceding assumptions allow us to set the first element of the
shadow wage rate for formal sector workers, M., equal to the value at
accounting prices of the marginal product of labor in agriculture or
unemployment. The value of the marginal product of an unemployed worker is
taken as equal to zero. Estimating the accounting value of the loss in
agricultural output requires two parameter values, an estimate of the marginal
value product at market prices of agricultural labor and an accounting ratio
applicable to agricultural output.
There is an active market for hired labor in the Egyptian rural
economy with a considerable proportion of rural households being dependent
upon earnings from wage labor and rural off farm employment for their
livelihood. Table 4-9 summarizes the sources of income of rural households.
Column (4) of the table shows the proportion of the total number of households
in the sample which have an income originating from each of the nine sources
listed in column (1). It emerges that 48 percent of households in the sample
had income originating from family farms while 35 percent derived at least a
portion of their total income from agricultural wage employment. Particularly
striking is the importance of off farm employment as a source of income with
19 and 30 percent of households reporting sources of income from village and
non-village off farm employment respectively.
I/ We find in the Appendix that our estimate of the shadow wage may overstate
the "true" shadow wage by 5-7 percent if wages are bid up.
- 174 -
Table 4-9: Distribution of Rural Household"Income Claims" by Source
No. of % of % of all HouseholdIncome Distribution Having an IncomeClaims Claims Claim from Each Source
1. Wage other farms 305 17.9 35.22. Wages off farm in
village 163 9.6 18.83. Wages outside village 261 15.3 30.14. Family farm with
labor input 418 24.5 48.25. Family farm without
labor input 185 10.8 21.36. Rent of land 54 3.2 6.27. Rent of equipment
Note: These data were generated from a survey carried out by the ILO in 1977.The Survey covered a sample of 1,000 households liviing in 18 villages.For details see Samir Radwan and Eddy Lee,-The Anatomy of Rural Poverty:Egypt, 1977, 1980.
A recent ILO (1980) report argues-that the rural labor market is
becoming increasingly important and competitive as traditional institutional
arrangements, sexual differentiation of tasks and cultivation practices are
abandoned in response to changing economic incentives and technologies. Thus
it would seem reasonable to use the typical wage rate for hired agricultural
labor as an estimate of labor's marginal value product. Seasonal fluctuations
in labor demand, however, result in substantial variations in daily wage rates
and periods of employment throughout the year, and we must make some allowance
- 175 -
for the high incidence of non-farm employment as a source of income since
withdrawing a typical agricultural worker from the rural economy will also
result in a decline in off farm output.
Mohie-Eldin (1980) has estimated that, taking days of employment and
wages throughout the crop cycle, the average annual income from one farm wage
employment by male workers was LE 125 in 1976. To this he adds 20 percent for
off farm employment and arrives at an estimate of the annual income of male
agricultural wage laborers of LE 150. We convert this to 1975 prices using
the rural consumer price index and arrive at an estimate of the value of
output foregone by withdrawing an agricultural worker from employment of
LE 129 at constant 1975 prices.
The cost of withdrawing labor from rural employment in terms of the
unit of account is computed on the assumption that wage labor is withdrawn
from alternative activities in proportion to its employment. The relevant
calculation is set out in Table 4-10. Since the farmgate price lies below the
international market price for several important commodities, the weighted
ratio of output foregone at accounting prices to output foregone at market
prices is approximately 1.18.
The social value of the marginal product of an agricultural wage
laborer is given by:
M = LF 129 (1.18) = 152.22a
at accounting prices which we round down to LE 152.
Given the assumptions outlined above, the only other sector from
which a worker is withdrawn without replacement and, therefore, where there is
possibly some decline in output is in the public industrial sector. Redundant
workers may have a positive marginal value product despite the fact that it is
- 176 -
exceeded by the wage. At this point we shall make the extreme assumption that
the marginal product of a redundant worker in public industrial enterprises is
zero. (Indeed some commentators have argued that the marginal value product
of such workers is negative). We may, however, modify this assumption in
subsequent work when we have an empirircal basis for estimating the marginal
productivity of labor in public enterprise.
Table 4-10: A Conversion Ratio for Agricultural Production
Activity Labor Input C-opping Weights Accounting(Man Days Per Feddan) Pattern Ratio
(b) The Resource Cost of Workers Transferring Occupations
When a worker moves from other employment (or unemployment) into a
formal sector job, his household income will rise by the excess of formal
sector wages over wages in his previous employment adjusted for any change in
the earnings of other household members. For example, when an agricultural
worker migrates to an urban formal sector job, his disposable income increases
by the excess of formal sector wages over his earnings in agriculture. If he
migrates with his family, however, household income may not increase by the
- 177 -
whole of the difference, should earnings of other members decline as a result
of their move to the city. The second element of the shadow wage is the cost
in terms of the unit of account of providing the resources to meet the
increase in expenditure and savings brought about by an increase in household
income as workers are drawn into formal sector employment. It may be defined
using the following notation:
YFi = household income in final employment in i;
i = u for urban employment, i = r for rural
employment.
YAi = household income in alternative employment in
area i.
s marginal propensity to save assumed constant
for all households.
fs = accounting ratio for savings of workers'
households.
fci accounting ratio for marginal increases in
consumption of low income households in sector
1.
fci = accounting ratio for average consumption of
low income households in sector i.
P. = the share of household income remitted toiJ
relatives in sector j from sector i.
P. = the value of contributions by relatives iniJ
sector i to the support of unemployed workers
in sector j.
- 178 -
It is possible to present two general formulae for the resource cost
of workers changing occupations depending on whether the worker and his
household migrate from one area to another to take up final employment.
Consider first the simple case of a worker changing occupations without
migrating. A general formula for the resource of his transfer is;
CFiAi = (YFi -yAi )(sf + (l -s)f i) (2)
If the worker and his family migrate from one area to another to take up
employment, the household may face altered relative prices for consumption
goods. In such circumstances, patterns of expenditure will differ between
locations and the accounting ratio applicable to pre- and post-migration
expenditure on consumption will differ. The resource cost for migrants
changing occupations is given by:
FiAj PijYFi(sfs + (1 - s)f .)
+ ((1 - Pij)YFi YAj) s
+ (1 - s)((l - pi j)YFi ci - yAj cj
If we make the additional simplifying assumption that urban relatives support
the urban unemployed and rural relatives support the rural unemployed, these
general formulae also apply to cases of workers drawn from or entering into
the pool of unemployed. 1/
l/ Consider for example a worker drawn into employment from the unemployedwho does not migrate. While he was unemployed relatives contributedsupport Pi The resource cost of the change in income of the workerand his relatives is given by;
thus receipts from relatives and expenditure by the unemployed cancel outand the resource cost is given by the accounting value of the consumptionand savings out of the wage in employment.
- 179 -
Estimates of the resource cost of workers changing occupations are
contained in Table 4-11. The entries in Column (1) refer to a male head of
household engaged in the occupations listed. For each type of activity other
household income is presented in Column (2). This is then divided between
income accruing to rural and urban households and between consumption and
savings. Market values are converted to numeraire values using the conversion
ratios for consumption and savings. The resource cost of a worker changing
occupations is given by the difference between the resource costs of the two
occupations required. 1/
The assumed value of the marginal propensity to save is singularly
uncertain. For convenience we assume that the marginal propensity to save of
all households is 10 percent. The national household budget survey does not
make any effort to examine savings behavior of households. The 10 percent
value for the marginal propensity to save is similar to that found for example
in India and Kenya but it cannot be justified on any empirical basis in Egypt.
All incomes and transfers of income in Table 4-11 are expressed in
constant 1975 prices. We have previously discussed the origin of our estimate
of the annual income of a fully employed wage laborer in agriculture. The
earnings of the rest of the family are based on the same 1977 ILO survey for
household of the size and composition appropriate to agricultural wage
employment. In evaluating the household incomes of the rural and urban
1/ Strictly this may only be done because we have been forced to use marginaland average accounting ratios for consumption which are the same. On this
point, see M. FG. Scott, J. D. McArthur, and D. M. G. Newbery, (1976).
Table 4-11: The Resource Cost of Unskilled Workers Changing Occupations
Head of Other Transfers Rural Urban Social CostHousehold Members From Relatives Savings Consumption Saving Consumption Saving Consumption
unemployed we have assumed that there is no change in the earnings of other
family members. The level of transfers from family members outside the
household was estimated by taking the household income per capita necessary to
sustain minimum subsistence (ILO (1980)) and multiplying by the number of
members in the household at that level of income. The difference between this
subsistence household income and the earnings of other family members was
assumed to consist of transfers from relatives. Finally, we consider the
basis for our estimates of the earnings of fully employed unskilled laborers
in the formal sector. Earnings of rural formal sector workers were taken from
the annual earnings of off farm non-village employment from the 1977 ILO
survey, converted to 1975 prices by the rural CPI. The annual wage for urban
public sector employees is taken from Handoussa (1980c) which gives average
earnings of production workers in the textile sector (most of whom are largely
unskilled) for 1970-1978. The average wage was adjusted downwards by 25
percent to allow for the effects of seniority and skill composition, and the
entry in Table 4-11 is the average at constant 1975 prices of this adjusted
wage for the period 1976-1978. The private sector wage was found by taking
the average of wages for unskilled construction workers and for production
employees of private sector manufacturing establishments employing 10 or more
workers. The annual wage estimated in this way is very close to the adjusted
wage in the public sector and provides a measure of corroboration of the two
independently derived estimates.
Earnings of the remaining members of the household were estimated by
taking the difference between the average household income of families headed
by production workers and the estimated earnings of the head of the
household. The predicted level of earnings of other household members derived
- 182 -
in this way is quite close to that of a rural household. If correct, this
indicates that there is little sacrifice of earnings by other family members
resulting from the migration of an unskilled worker from rural to urban areas
in Egypt. 1/
(c) Benefits from Increases in Income of
Workers Changing Occupations
When a worker moves from employment in one sector to another, the
social gain (or loss) resulting from his change of occupations is found by
evaluating the income gains and losses of all households affected by the
change of occupation. For example, if an agricultural worker moving into
formal sector employment remits nothing to relatives, all gains in income
accrue to the worker and his household.
The various elements involved in estimating the benefits to the
worker and his relatives from the change in occupations are summarized in
Table 4-12. The entry for the "compensation factor" is the pecuniary value of
compensation for a change in the conditions of work as a result of changing
occupations. For example, in line 4 the worker is assumed to migrate with his
family to the urban sector. He remits nothing to family in the rural sector
but he and his household must pay urban prices. In Table A3-1 the rural urban
1/ This may not be an unreasonable conclusion for Egypt where there isevidence of "circular migration" of households into and out of urban areasduring the year. Because of the relatively compact spatial organizationof the country, it is possible for households to engage simultaneously inboth urban employment an agricultural employment (ILO (1980)).
- 183 -
price differential for Egypt was estimated to be 10 percent in favor of urban
areas. Thus, the worker gains real income at rural prices of LE 33 due to the
lower urban prices. But there is the possibility that the movement of the
household from rural to urban employment may entail costs of migration. Given
our uncertainty concerning the magnitude (and even the direction) of the rural
urban price differential, we have chosen to consider that the gains in real
income are offset by other costs of migration and, therefore, that there is no
pecuniary value of the change in conditions of work in moving from
agricultural to urban formal sector employment.
In lines 1 and 2 the worker is assumed to leave agricultural
employment and enter the rural or urban unemployed. In such cases, relatives
are assumed to provide subsistence income of I.E 10 per year in rural areas and
LE 16 per year in urban areas. Urban relatives are assumed to support the
urban unemployed and rural relatives to support the rural unemployed. Thus,
the worker and his family sacrifice disposable income equal to LE 119 in rural
areas or LE 113 in urban areas as a consequence of the head of the household
becoming voluntarily unemployed. The compensation factor as a result of the
change in conditions of work is set equal to the loss in disposable income,
implying that the net loss incurred by the worker as a result of his movement
into unemployment is zero. The assumption of zero sacrifice may be justified
along the following lines. If the worker is free to remain in agriculture,
yet he chooses to be unemployed, he may not feel that he is incurring any loss
in welfare. His loss in income must then be exactly compensated by an
increase in leisure and mobility which permits him to seek alternative
employment, and the net loss resulting from the change in occupations is
zero. Alternatively, it is possible to allow for the willingness of a worker
- 184 -
to trade current income for the increased probability of finding a formal
sector job and hence a rise in expected future income. It seems plausible to
argue that entering rural unemployment will not materially increase the
probability of finding a formal sector job, and thus the worker should not be
prepared to make a large pecuniary sacrifice. For workers transferring to the
urban unemployed, the chance of finding a formal sector job may improve.
Thus, they may be prepared to incur some monetary loss in order to achieve an
expected increase in income. Any adjustment, however, would be essentially
arbitrary, and we shall not pursue the point further.
The social benefit in terms of the numeraire of the change in
disposible income is given under two sets of assumptions. In column (7), we
present the benefit appropriate to the calculation of the extended efficiency
wage which from Section 4.2 is b = 1/v times the change in income. In column
(8), we consider the distributional gain associated with the set of welfare
weights derived in Chapter 3 on the assumption that n = 1. We evaluate the
welfare function in Section 3 at each income and compute the value of the
intramarginal change.
It is possible to use the table to determine the net benefit of
unemployed workers moving into alternative employment. For example, consider
a member of the rural unemployed who moves into a rural formal sector job.
The net social benefit from the transfer of occupations is assumed to be given
by the difference between the net gains shown in lines 1 and 3 and Table
4-12. This is strictly correct for marginal workers since the form of the
utility function is such that if the benefits of moving from YO to YI and
from YO to Y2 are known the benefit of moving from YI to Y2 can be
deduced.
r 185 -
Estimating the Proportion of Workers Drawn from Different Sectors
In order to draw the estimates of social cost for each of the cases
discussed into a single estimate of the shadow wage, some measure of the
probability of drawing a worker from each of the cases examined must be
constructed. To summarize the relevant simplifying assumptions; workers are
assumed to be drawn from the agricultural sector, or from the rural or urban
unemployed. Drawing a worker from the casual sector or out of other formal
sector employment is assumed to have no effect on employment or wages in
either sector, and workers drawn from these sectors may be regarded as having
been indirectly drawn from agriculture or the unemployed. Estimates of the
proportion of formal sector workers drawn from agriculture and unemployment
are presented in Table 4-13.
As a first approximation, it was assumed that workers are recruited
into the formal sector in direct proportion to their numbers in the pool of
eligible labor -- represented by the total of agricultural employment, and
rural plus urban unemployment. Such an assumption is consistent with a
situation in which the creation of a new formal sector job results in complete
transmission of information regarding the job opening to all sectors. Workers
in all sectors are assumed to have equal opportunities to obtain the job.
In Column (2) of the table we allow for the possibility that the
creation of a new formal sector job may result in the release of some
redundant labor from existing public sector enterprises. There is no
objective method to establish the probability that a worker will be drawn from
existing redundancies in the public sector. We have, therefore, made the
assumption that redundant workers are released in proportion to their number
in urban employment. As a best estimate of the number of redundant workers in
the public sector we take 20 percent of the public sector labor force.
- 186 -
Table 4-13: Proportions of Workers Drawn intoFormal Sector Employment
The final estimates in Table 4-13 are based on applications of the
Harris-Todaro model of rural-urban migration. Estimate 3 makes use of the
simplest version of the model. 1/ Job seekers among the urban unemployed
equate the expected urban wage to the wage in agriculture, W = pW , whereau
p is the probability of obtaining a formal sector job. In the model
p = Lf/(Lf + L ), where Lf is the labor force in formal sector urban
employment and L is the number of urban unemployed. If an additional urban
1/ J. R. Harris and M. P. Tadaro, "Migration Unemployment and Development: ATwo Sector Analysis," American Economic Review, March 1970. For a model
based on the "small country assumption" of given world commodity pricessee W. M. Corden and R. Findlay, "Urban Unemployment, IntersectoralCapital Mobility and Development Policy," Economica, 1974.
- 187 -
sector job is created and wages remain constant in both agriculture and the
formal sector, p must remain constant and dL IdLf = L ILf. The
estimates of urban employment and unemployment in Table 4-13, permit a rough
estimate of dL IdLf of .119. Thus in estimate 3 urban unemployment is
given a negative weight of -.119.
Although the simple Harris-Todaro model makes it appear that an
increase in urban unemployment is inevitable, Scott has argued convincingly
that in the context of a dynamic model of rural-urban migration there are no a
priori grounds to assume that unemployment will either increase or decrease as
a result of increased formal sector employment. l/ Indeed, the limited
empirical work available does not lend strong support to the Harris-Todaro
hypothesis in its extreme form. Attempts to correlate the unemployment rate
in a number of LDC's to formal sector vacancies have been largely
unsuccessful. This suggests that in Egypt it is conceivable that
dL /dL = 0. In column (4), the proportion of unemployed workers is setu f
equal to zero. Thus all workers entering formal sector employment are assumed
to originate in agriculture. Corden and Findlay have noted that the
assumptions underlying estimates 3 and 4 form limiting cases of the more
general Harris Todaro type model W = apW where a is a parameter
reflecting the average level of risk aversion of migrants. Thus, if the
Harris-Todaro model is applicable to Egypt the limits of the impact on urban
unemployment should be given by estimates 3 and 4.
1/ Scott, McArthur and Newbery, (1976).
- 188 -
Our assumptions regarding the sources of rural formal sector
employment are similar. In columns (1) and (2), we employ the same
assumptions and proportions as in the case of urban employment. In column
(3), we allow for the possibility that rural jobs are filled only from the
rural economy and in column (4), we assume that they are filled wholly from
agriculture.
(d) Shadow Wages for Workers Drawn into the Formal Sector
Tables 4-14 and 4-15 summarize our estimates of the social cost of
labor drawn into formal sector employment from the four sources in Table
4-13. Four weighted averages are computed on the basis of the weights
provided in Section 4.4(c) above. Thus the tables provide a sensitivity
analysis of the shadow wage with regard to the form of market adjustment
assumed to take place in the labor market.
Our estimates of the shadow wage are presented for each of the three
accounting price methods discussed in Section 4.2. At efficiency prices, the
urban shadow wage for unskilled labor lies in the range of LE 117-168. Social
accounting prices yield a similar magnitude for the shadow wage of LE 104-141,
indicating that given our system of welfare weights the benefits associated
with the increase in household consumption made possible by the change in
employment slightly more than offset the costs. At extended efficiency prices
the shadow wage is substantially higher, in the range from LE 204 to 276. The
consumption costs of the increase in household income weigh more in the
extended efficiency price system because of the premium on public sector
income relative to consumption by individuals at all levels of income.
Our estimates of the rural shadow wage are qualitatively similar to
those for urban unskilled labor, although the estimates of social cost at both
Table 4-14: Shadow Wage Rates for Urban Unskilled Labor
Sources of Labor Components of Social Cost Social Cost of Worker
Mj Cj B- Efficiency Extended Efficiency SocialOEP SAP Prices Prices Prices
Agriculture 152 214 118 239 152 248 127
Rural Unemployment 0 334 170 258 0 164 76
Urban Unemployment 0 332 175 257 0 157 75
Public Sector Employment 0 0 0 0 0 0 0
(Redundancies)
Average Social Cost Weighted by Column I weights 138 240 122
Proportions of Workers from Table 4 (.420) (.729) . (.372)
Table 4-15: Shadow Wage Rates for Rural Unskilled Labor
Sources of Labor Components of Social Cost Social Cost of WorkerMj Cj Bj Efficiency Extended Efficiency Social
EP SAP Prices Prices Prices
Agriculture 152 143 118 233 152 177 62
Rural Unemployment 0 262 126 252 0 136 10
Urban Unemployment 0 260 131 251 0 129 9 1
Public Sector Redundancies 0 0 0 0 0 0 0 °
Average Social Cost.Weighted by Column 1 weights 138 173 57Proportions of Workers from Table 4 (.515) (.646) (.213)
Column 2 weights 117 147 49(.437) (.549) (.183)
Column 3 weights 150 176 61(.560) (.657) (.228)
Column 4 weights 152 177 62(.567) (.661) (.231)
- 191 -
extended efficiency and social accounting prices are substantially lower
reflecting the lower assumed level of rural formal sector wages and (in the
case of social accounting prices) the higher welfare weight attached to the
change in income. Indeed our estimates of the shadow wage at social
accounting prices are quite low because of the welfare benefits of raising
rural household incomes.
All of our estimates of the shadow wage fail to allow for the
possible effect of bidding up of wages in the agricultural labor market and
are therefore subject to a measure of upward bias. Our best estimate of the
shadow wage and the ratio of the shadow wage to the market wage is presented
in Table 4-16.
Table 4-16; Estimated Shadow Wages and Accounting Ratios
for Rural and Urban Unskilled Labor
Efficiency Extended Efficiency Social Accounting
Prices Prices Prices
Urban Unskilled Labor 150 250 130
(.456) (.760) (.395)
Rural Unskilled Labor 150 175 60
(.560) (.653) (.224)
These shadow wage rates apply to formal sector projects in the public sector
and the "organized" private sector (firms with more than 10 employees) only.
Our results indicate that in these sectors of the urban economy the shadow
wage for unskilled labor is about 40 percent of the market wage at both
efficiency and social accounting prices and about 75 percent of the market
wage at extended efficiency prices.
- 192 -
Appendix 4-1: Formulae for the Shadow Wage When Wages are Bid Upand Test for Biases in the Estimates of Shadow Wages
The accounting prices for rural and urban unskilled labor estimated
in Section 4.4 are based on the assumption that an increase in vacancies in
the formal sector will not result in any bidding up of wages in the sectors
from which new formal sector employees are drawn. In this appendix we develop
a more general formula for the social cost of labor when wages are bid up and
attempt to estimate the probable direction and magnitude of the biases
introduced by ignoring the bidding up of wages in our estimates of shadow
wages.
Initially, we consider a two sector economy. Labor in the urban
formal sector is paid an institutionally fixed minimum wage Wf. Labor in
the alternative sector of the economy, which is assumed to be agriculture is
paid a competitively determined wage, Wa) which is less than the formal
sector wage. There is no urban unemployment, and "Harris-Todaro" effects are
non-existant. 1/ A new formal sector project is established which withdraws
labor from the agricultural sector equal to the amount QlQl in Figure A.
Wages are bid up in the agricultural labor market to W which in the diagrama
results in a net fall in the services of agricultural labor market of
QlQl- Thus, as we have presented the rural labor market there is some
elasticity of the supply of effort in agriculture which results from the
l/ Actually these assumptions are slightly more restrictive than necessaryfor the simple model. We could have assumed that the Harris-Todaro modelworks as in case 4 in Section 4.4, and that the pool of urban unemployedremains constant in which case the estimate of the social cost of a workerdrawn into the formal sector is analytically equivalent to the onepresented above.
- 193 -
Uw.
t; , q, (ig
Figure 4-1
'Impact of an Increase in Wages on the Labor Market
- 194 -
remaining workers increasing their hours of work or from drawing new
participants -- wives and children, for example -- into the rural labor force.
The general formula for the social cost of labor drawn into the
formal sector from agriculture was given in Section 4:2
SWR = M + C - B (a)af a a a
In this case, however, the elements of the shadow wage themselves consist of
two components, the direct costs and benefits arising from the movement of
workers dQ = QlQl plus the indirect costs and benefits of changes in
output and effort in the rural labor market as a result of the rise in wages.
We introduce the following notation to supplement that of Section 4.4:
dQd = The decline in demand for agricultural workers
remaining in the rural sector as a result of the
rise in wages, e.g., QlQl in Figure A.
dQ = The increase in the supply of effort from workers
remaining in the agricultural sector, e.g.,
dW The rise in the market wage of agricultural labor
resulting from the bidding up of wages in the
sector, e.g., W W.a a
We shall further assume that only the incomes of heads of households
are affected by a change in occupation or wages. Thus, we derive expressions
for each element of the shadow wage. If agricultural workers are paid the
value of their marginal product at market prices, the market value of the
decline in output is given by the decline in demand for agricultural labor
multiplied by the wage in the pre-project equilibrium.
W dQd = Ql- dW na (b)
where the elasticity of demand for labor in the agricultural sector is defined
- 195 -
as customary as:
W dQd (c)
a Ql dW
The social valuation of the loss in output is, therefore:
Ma = fa a Q dW * n (d)
where f is the accounting ratio applicable to the marginal product of labora
in agriculture.
The increase in effort on the part of workers remaining in the
agricultural sector increases their disposable incomes by:
Wa* dQ = Ql- dWa g (e)
where the elasticity of the supply of effort by workers in agriculture is
defined as:
W dQ (f)Oa Q - dW
1 a
The value of the social cost imposed by providing the resources to meet the
increased consumption and savings by workers remaining in the agricultural
sector is: 1/
fyr- Q1 dWa ga (g)
The increase in disposable incomes of workers transferring from agricultural
employment to formal sector employment is:
dQp(Wf - Wa) (h)
and the social cost of providing the resources to meet this increase in
1/ fyr is the conversion factor applicable to marginal increases in ruralincomes (cfcr + sf5). Because the conversion ratios for rural andurban consumption are nearly identical we use this also to apply toincomes of agriculturalists moving to urban formal sector employment.
- 196 -
consumption and savings is:
fy * dQp(Wf -wa) (i)
Finally, we consider the benefits arising from the increases in
income. We assume that the increase in income of workers in the agricultural
sector exactly compensates them for the disutility of increased effort, and
thus there is no net benefit accruing from the rise in disposable income as a
result of increased effort. The rise in wages, however, also introduces a
transfer of income from employers of agricultural labor and, perhaps,
ultimately from consumers of their products to workers in the agricultural
sector. The value at market prices of this transfer of income is:
Ql. dW (j)
The social valuation placed on the transfer of income is given by;
Ql dW- (baw - bae) (k)
where baw and bae are the welfare weights given to marginal changes in the
income of agricultural workers and employers (or consumers of agricultural
output) respectively. These weights are determined in the manner described in
Section 3.7.
The social valuation placed on the gain in income to workers drawn
from agriculture into formal sector employment is determined in the manner set
out in Section 4.4 for non-marginal changes in income. It may be represented
here as:
dQ (W - W)b (l)p f a af
where baf is the weight determined on the basis of the formula for
non-marginal changes in income for typical agricultural workers transferring
from the agricultural to the formal sector.
- 197 -
The complete expression for the social cost of transferring workers
dQ from agriculture into formal sector employment is given by:
Q dW nafa + dQ (W f-W a)fyr + Q *dW *g. fyr
- dQ (W -W ) b -QdW (b -b ) (i)p f a af a ae aw
rearranging terms and noting that QdW = dQ W/(n + g ) we derive:
dQp(wf-WaXf yrf fb baf)
dQ W (n f + gf (b - b)) (n)+ -p a a a ayT *aw ae
(n + ga a
setting dQp = 1, the shadow wage rate for workers drawn from agriculture
when wages are bid up is:
SWR = (W -W )(f -b )af f a yr af
Wa(naof + g a*f -(b -b )) (a)+a a a a yr aw ae (
(nr + g )a a
Ignoring the effect of rises in agricultural wages in Chapter 4
allowed us to employ a much simpler version of the shadow wage rate. Because
we were able to assume that the market value of the fall in output was equal
to dQ W,a and because there were no indirect effects on the rural labor
market, the simplified form of the shadow wage was given by:
NWR = W= f + (W -W )(f -b ) (p)af a a f a yr af
Under what circumstances will the two measures approximately coincide and what
are the biases introduced into the estimates of the shadow wage by use of the
simplified form in Chapter 4? If wages are in fact bid up, we define the bias
as:
afR SWR Waf _ Wanafa _gafya f (b -b )) (q)
afa +g)
- 198 -
The formula in Chapter 4, SWR fl will equal the true shadow wage in the
presence of a rise in agricultural wages, if the elasticity of the supply of
effort in the agricultural sector is zero, and the net benefit from the
transfer of income from employers to employees (b -b ) is also zero. It
does not seem unreasonable to assume that for marginal changes in agricultural
employment and wages the supply curve of labor services is highly inelastic.
Increases in effort are limited by available working hours and by preferences
for other activities. It is commonly argued that subsistance farmers exhibit
a marked preference for leisure, and while it is beyond the scope of the
present paper to review the literature on labor supply response in traditional
agriculture low or zero elasticities of supply are not inconsistent with the
models presented. 1/
It is interesting to note that the case of a backward bending supply
curve of labor can also be handled within the general framework for the shadow
wage presented. In this case, which has been widely discussed in the
literature on the supply of effort in peasant agriculture, the elasticity of
supply of labor is negative. Reference to the general formula SWR f
indicates that in such a case the indirect effect resulting frm the bidding up
of wages will involve a fall in output greater than the value W dQ at
market prices. Set against this will be a decline in the resources needed to
1/ See for example, the contributions by C. Nakajima and J. W. Mellor in
C. R. Wharton, Jr., Subsistence Agriculture and Economic Development
(London, Frank Cass, 1970). In the simplest case of all labor in wage
employment and a utility function in labor income and leisure, the shape
of the supply curve for agricultural labor depends on the nature of
preferences of workers. If both income and leisure are normal goods a
completely price inelastic supply curve of labor can be obtained without
extreme assumptions regarding the shape of individual preference maps.
- 199 -
support the consumption and savings of the rural labor force. The net result
will in general be that given a backward bending supply curve of labor the
shadow wage rate will exceed that for forward falling supply curves.
Note also that as the elasticity of supply approaches infinity the
estimated bias approaches W f -W f , regardless of the elasticity ofa a a yr
labor demand. This is because the increase in effort by remaining workers
offsets the decline in output caused by removing one worker from the
agricultural sector, but at a cost in terms of the additional resources needed
to provide for the extra consumption of those remaining. If the accounting
ratios for agricultural output and for consumption by agriculturalists are
approximately equal, as they are in Egypt, the simplified formula for the
shadow wage which we have used in Chapter 4 is quite a good estimate of the
social opportunity cost of labor.
In order for b to equal ba the incomes per capita in the
household of employers and employees must be approximately equal. 1/ To the
extent that the transfer of income is from farmers in their capacity as
employers to farmers and their families in their capacity as workers the
social benefit is nil. Not all of the agricultural labor force is
self-employed, however, and if on average workers are less well off than their
employers this portion of the transfer of income will represent a net social
1/ The manner in which welfare weights are derived from information on incomeper capita is presented in Section 3.7. It should also be noted that ifthe rise in wages were passed wholly on to the consumers of agriculturalproducts the transfer of income would be from consumer to workers. Becauseprices for many Egyptian agricultural commodities are administrativelyset, this was not felt to be a major consideration. To the extent thatthe rise is passed on in higher prices the transfer of income willincrease the benefit somewhat more than the case presented in the text.
- 200 -
benefit. Data on wage employment in agriculture are severely limited. In
Chapter 4 we arrived at an estimate of the average share of the rural labor
force in wage employment in agriculture of 15 percent. If we further assume
that all agricultural laborers belong to the LE 150-199 income class in Table
3-15 and that employers belong to all households taken at random, the benefit
resulting from the transfer of income will be equal to
Wa (2.3 3-1.87)(.150)/ ri, for an elasticity of the marginal valuation of
income of one. Thus, for elasticities of demand for agricultural labor equal
to one or greater than one, the benefit arising from the transfer of income
will be less than 7 percent of the agricultural wage.
Table A4-1 presents a sensitivity analysis of the estimated bias in
the shadow wage caused by our failure to consider the possibility of bidding
up of agricultural wages. Parameter values underlying the table reflect
estimates in Chapter 4 of the agricultural wage, the conversion ratios for
agricultural output and consumption, and the marginal benefits assigned to the
income transfer. The values in the table are the percentage by which the
estimate of the social accounting cost of labor drawn from agriculture in
Chapter 4 of LE 127 at accounting prices exceeds the true social opportunity
cost when wages are bid up. To interpret the table consider two extreme
cases. If the elasticity of labor supply is zero, the estimated resource cost
in Chapter 4 results in an overstatement of the social cost of labor obtained
from the agricultural sector of about 7 percent due to the welfare benefit of
the transfer (recall that the increased cost of consumption by workers is
precisely offset by the reduced consumption of employers). In the other
extreme, if the elasticity of the supply of effort is infinite the estimated
resource cost also overstates the opportunity cost, but only by about 10
- 201 -
percent. This is because, althought there is no decline in output, additional
resources are required to provide for the increased consumption of the
remaining workers in agriculture.
If we take as our best guess that the estimated resource cost of the
marginal worker drawn from agriculture overstates true opportunity cost by 65
percent, this implies that our estimates of shadow wages at social accounting
prices overstate the true opportunity cost of labor by between 5-7 percent
depending on the weighting scheme used. This is well within the confidence
interval around our point estimate of the shadow wage.
Table A4-1: Estimated Biases in the Shadow WageWhen Rural Wages are Bid Up
Elasticity of Labor Supply Elasticity of Labor Demand in Agriculture (na)(ga) na =0 na =1 na = 2 na = 5
ga = * 7.0 3.4 1.4
ga = 1 17.1 8.6 6.7 2.8
ga = 2 13.7 9.1 6.8 4.0
ga = 5 11.5 9.3 6.3 5.3
ga = 10.1 10.1 10.1 10.1
Notesz Wa = 129
baw = 2.33bae = 1.87
fa = 1.18
fya = 1.08
- 202 -
References
Cuddihy, William (1980) Agricultural Price Management in Egypt, World Bank
Staff Working Paper No. 388, April.
Dervis K., R. Martin, and S. van Wijnbergen (April 1982) An Optimal Growth
Model for Egypt: Issues in Investment Allocation, Shadow Pricing and Oil
Extraction, mimeographed, World Bank.
Edil, Reda el (1980) "Impact of Taxation on Income Distribution - An
Exploratory Attempt to Estimate Tax Incidence in Egypt" in G. Abdel-Kalik
and R. Tigmor (edns.) The Political Economy of Income Distribution in
Egypt, New York: Holmes and Meier (forthcoming).
Handoussa, Heba (1980a) "The Impact of Economic Liberalization on the
Performance of Egypt's Public Sector Industry", paper presented to the
Second B.A.P.E.G. Conference, April 2-5.
(1980b) "Liquidity Problems of Public Sector Enterprises", Cairo.
(1980c) "Public Sector Employment and Productivity in theEgyptian Economy", Report Prepared for the ILO Comprehensive Employment
Strategy Mission, September.
I.L.O. (1980) Employment Opportunities and Poverty in a Changing Economy:
Egypt in the 1980's. Report of the ILO/UNDP Employment Strategy Mission,
June-October.
Korayem, Karima (1980) "The Agricultural Output Pricing Policy and the
Implicit Taxation of Agricultural Income" in G. Abdel-Kalek and R. Tigmor
(edns.). The Political Economy of Income Distribution in Egypt, New York:
Holmes and Meier (forthcoming).
Little, I.M.D. and J. A. Mirrlees (1974) Project Approval and Planning for
Study, Report to Egyptian General Petroleum Corporation.
- 203 -
Ray, A. (1980) Comversion Factors and Shadow Exchange Rates, World Bank:Central Projects Note No.2.03, July.
Sabot, R., Taylor, L. and Boutros-Ghali, "Labor Force Growth, Employment andEarnings in Egypt: 1966-1986" World Bank Population and Human ResourcesDivision Discussion Paper Bl-16.
Schohl, W.W. (1979) Estimating Shadow Prices for Columbia in an Input-OutputTable Framework, World Bank Staff Working Paper No.357, September.
Scott, M. FG. (1977) "How to Use and Estimate Shadow Exchange Rates" OxfordEconomic Papers, Vol. 26, No.2, July, pp 169-84.
Scott, M. FG., D. MacArthur and D. M. G. Newbery (1976) Project Appraisal inPractice, London, Heineman Educational Books.
Squire, Lyn, I.M.D. Little and M. Durdag (1979) Application of Shadow Pricingto Country Economic Analysis with an Illustration from Pakistan, WorldBank Staff Working Paper, No.330.
Squire, L. and M. H. van der Tak (1975) Economic Analysis of ProjectsBaltimore, The John Hopkins University Press.
Werner International (1976) Sector Survey of the Egyptian Textiles Industry,Report submitted to the Egyptian Central Organization for Spinning andWeaving, Cairo.
World Bank (1980a) Arab Republic of Egypt: Domestic Resource Mobilization andGrowth Prospects for the 1980's, World Bank Report No.3123-EGT,December 10.
World Bank (1980b) Egypt: Economic Management in a Period of Transition: AWorld Bank Country Economic Report, Baltimore: The John Hopkins UniversityPress.
World Bank (1981a) Arab Republic of Egypt: Agriculture Price Policies Study,July.
World Bank (1981b) Egypt's Past Performance in Real Capital Accommodation(1974-79).
The World BankHeadquarters:1818 H Street, N.W.Washington, D.C. 20433, U.S.A.