You are on page 1of 16

Project Appraisal

ISSN: 0268-8867 (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tiap18

Conversion factors and shadow exchange rates

Gordon Hughes

To cite this article: Gordon Hughes (1986) Conversion factors and shadow exchange rates,
Project Appraisal, 1:2, 106-120, DOI: 10.1080/02688867.1986.9726548

To link to this article: https://doi.org/10.1080/02688867.1986.9726548

Published online: 17 Feb 2012.

Submit your article to this journal

Article views: 5711

View related articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tiap18
Project Appraisal. June 1986. volume I , number 2, paces 106-1.20. & w h Tree Piihlrrhin~.10 U"ur/ord C l o s ~ .Giuldjord, .Sirrre.v GUI zEP, England

Basic principles
Conversion factors and shadow
exchange rates

Gordon Hughes

OR SOME TIME the contents of various


In calculating shadow prices for use in project
appraisals, the rekrtionship between the shadow
exchange rate and converslon factors (accounting
F journals and discussion at meetings of development
economists have been punctuated by the crossfire between
ratios) for specific items or categories of inputs is a proponents of different methods of project evaluation. At
common source of confusion. This article sets out to the same time, some of the more important features of
clarify links between choice of numeraire, shadow these procedures - shadow wage rates, shadow exchange
exchange rates and conversionfactors in the UNIDO rates, social rates of discount - have been adopted
and Little - Mirrlees methodologies. It outlines a increasingly widely among those concerned with the
general method of estimating sets of shadow prices practical assessment of development projects.
which can easily be updated in the light of extra Understandably, many practitioners find themselves
information or as a result of changes in economic confused by the seemingly contradictory arguments of the
policy. protagonists. In particular, the divergent views expressed
Finully,the implications of an overvalued exchange by those who use shadow exchange rates and those
rate andor an impeding currency devaluation for working in terms of conversion factors have been
shadow pricing are &cussed. A pmedure to adjust troublesome.
shadow prices in such circumstcrnces is described. Thus, I embarked on this paper with the purpose of
clarifying the bases for these different views, so as to
facilitate the process of making the basic assumptions and
judgements which must underpin any social cost benefit
analysis.
In the course of thinking through the literature and
writing on this subject, it became clear that much of the
conflict arises because contributors to the discussion work
Keywords: shadow prices; social cost-benefit analysis; with different implicit models without ever really
exchange rates; devaluation elucidating the consequences of adopting alternative
assumptions. There is also a basic divergence between
those who tend to think in terms of aggregate or macro-
economic variables and those who derive their
Gordon Hughes is a Professor of Political Economy, recommendations from allocative planning models which
Department of Economics, University of Edinburgh, George focus on the micro-economic aspects of investment.
Square, Edinburgh EH8 9JY, Scotland. He has worked on the Thus, much of this paper is devoted to the careful
estimation of sets of shadow prices for project appraisal and exposition of some of the assumptions which underpin
policy analysis in several countries, including Indonesia, different methods of calculating shadow prices.
Morocco, Tanzania and Tunisia. In advocating the use of what I have called the
Little-Mirrlees-Scott (LMS) method of estimating shadow
prices, I should declare an interest in the sense that I have
had considerable experience of the practical
implementation of the method. It was originally developed
by Scott for use in Kenya and Mauritius - particularly
for the evaluation of the Million Acres Settlement schemes
in Kenya (Scott, 1976). Subsequently, I estimated sets of
accounting ratios for various countries including

106 0268-8867/86/02010&15 US$03.00 8 Beech Tree Publishing1986 Project Appraisal June 1986
Conversion factors and shadow exchange rates
Indonesia, Morocco and Tunisia (Hughes 1979, 1980) numeraire for project evaluation.
which have been used in the evaluation of both agricultural It seems fairly clear that the first best concept of the
and industrial projects. shadow exchange rate is not likely to be very helpful in
Certainly, on the basis of this work I can assert that the project evaluation, though it may yield useful information
method is practicable in that it does not require about the overall effects of the structure of trade
unreasonable amounts of data and a very extensive set of intervention when studying macro-economic policy in
accounting ratios can be estimated in the space of three some countries. While I hope that there would be no
month’s work, depending on the amount of preliminary disagreement about using a second best shadow exchange
work which has to be carried out. In addition, the method rate, it is unfortunately less clear which features of a
can be computerised quite easily, so that it is simple to particular structure of trade intervention should be taken
derive alternative sets of accounting ratios corresponding as given in deriving the shadow exchange rate.
to different assumptions about a range of fundamental In practice, the decision must rest on a judgement
parameters - such as income weights, shadow wages, the about which elements of the structure are likely to be
social value/cost of private profits. operative throughout the period during which the shadow
prices will be used. Generally my inclination would be to
assume that tax and tariff rates are fixed, but that
Basic issues additional quotas and other quantitive restrictions will or
should be eliminated. This is partly because such
The LMS method is outlined later. First, we will discuss a restrictions tend to be varied much more readily in
number of basic issues, the most important of which is the response to balance of payments pressures than tax or
choice of a numeraire and its effects on the appropriate tariff rates.
definition of shadow prices and shadow exchange rates. Thus, provided one allows for the appropriate exchange
Standard conversion factors are discussed next as there rate adjustment to achieve trade balance in the absence of
seems to be considerable confusion about their definition quantitive restrictions, it seems reasonable to assume that a
and use. In this paper I will use the term accounting ratio move towards more consistent and efficient government
to refer to the ratio of the shadow price to the market policies would involve the dismantling of quotas, import
price of a specific good on service. On the other hand, the licensing and similar schemes.
term conversion factor will refer to the multipliers used to Another reason for preferring this assumption is the
convert from market values to values at shadow prices for difficulty of making proper estimates of the social costs of
general categories of expenditure - consumption, non- the income transfers and other effects caused by
traded goods and services, and so on. quantitative restrictions on trade. The assumptions which
Finally, we consider an important practical problem in guarantee the equivalence between tariffs and quotas
the estimation of shadow prices which has not been very (Bhagwati 1969) are most unlikely to be satisfied in most
adequately discussed previously. This concerns the best Less Developed Countries (LDCs) so that the precise
way of modifying accounting ratios and shadow prices to effects of quantitative restrictions will depend upon the
allow for a prospective devaluation of the domestic way in which they are operated, on whether there is price
currency. Not only does this raise the issue of determining control for imported items, and so on.
the best method of incorporating the effects of the These contributory factors may well change over time,
prospective exchange rate change in our estimation so that it is better not to attempt to incorporate them in the
procedures, but also the question of the circumstances estimation of a general set of shadow prices. Rather they
under which the structure of relative shadow prices is should be specifically examined in those cases for which
unaffected by the change. the effects of a quota or other form of non-fiscal
There is one issue concerning both terminology and intervention may significantly affect the net social value of
assumptions which needs to be settled before we embark a project.
on any discussion of the calculation and interpretation of
shadow prices. In the literature, a distinction has been
drawn between first best and second best shadow exchange Shadow exchange rates and the numeraire
rates. Unfortunately these tend not to be defined very The choice of a numeraire should not affect the relative
carefully, especially as regards the numeraire, and the shadow prices derived for use in project appraisal so long
definitions tend to vary from author to author. as analysts adopt similar assumptions or simplifications in
The essential feature of any first best shadow exchange carrying out their work. However, much of the confusion
rate is that it is calculated on the assumption of optimal in the literature concerning the relationship between
trade policies - basically free trade, but allowing for alternative methods of project appraisal seems to flow
optimal export taxes or import duties when there are
inelastic foreign demand/supply functions. The choice of a numeraire should not
Second best shadow exchange rates assume that part or affect the relative shadow prices
all of the existing stucture of tariffs, quotas and other
forms of intervention should be taken as given, so that derived for use in project appraisal so
they indicate the social value of foreign exchange under long as analysts adopt similar
particular assumptions about trade and other policies. In assumptions or simplifications in
all cases I will assume that shadow exchange rates are carrying out their work
expressed in terms of aggregate consumption as the

Project Appraisal June 1986


Conversion factors and shadow exchange rates
from differences in implicit assumptions about government and it illustrates a general characteristic of procedures
behaviour and institutional arrangements prompted by which use shadow exchange rates or individual accounting
different choices of numeraire. ratios to perform the reverse process.
In this paper I will basically use the Little-Mitrlees The information requirements are broadly similar -
(L-M) numeraire - uncommitted income in the hands of since in the SER case we need to know the ratio of border
the government measured in terms of convertible foreign to domestic prices - but the process of estimating a single
exchange. As is usual, I will assume that the actual SER uses this data very inefficiently. Various
calculations are performed in terms of domestic currency assumptions are made to allow aggregation rather than
- rupees - with foreign (dollar) prices converted at the taking advantage of the opportunity to examine potentially
official exchange rate. important micro-economic aspects of economic policy and
This is, of course, based on the presumption that industrial structure which can be illuminated by the
border prices are the appropriate starting point in the estimaiion and study of individual accounting ratios.
derivation of shadow prices so that the process of Within the context of the UNIDO approach to project
estimating social costs and benefits is most conveniently appraisal, the exclusion of capital goods from the items
carried out in terms of border values. Note, however, that used in deriving the SER can be justified, but only at the
it does not imply that border prices themselves are always, expense of adopting assumptions which run counter to
or even usually, the correct shadow prices. other aspects of their philosophy. Their numeraire is
An alternative numeraire recommended by the UNIDO aggregate consumption while the SER is to be used to
manual is general consumption expenditure measured at convert uncommitted foreign exchange into units of the
domestic prices. It would appear that we can convert from numeraire.
this numeraire to the L-M one by multiplying all values Thus, the actual allocation of extra foreign exchange
by the Consumption Conversion Factor (CCF), while the between consumption and capital goods can be neglected
UNIDO approach requires the use of a shadow exchange by asking the question: if we have one extra unit of foreign
rate (SER) to convert foreign exchange items into values exchange, how many net additional units of aggregate
in terms of aggregate consumption. consumption can be provided after allowing for the
Hence, one might infer that the relevant SER is simply opportunity cost of any complementary resources?
the inverse of the CCF. This broadly corresponds to one In effect, the UNIDO procedure implies that the
version of the SER discussed in the UNIDO manual government utilises additional foreign exchange by making
(Dasgupta et a1 1972, Chapter 16 especially pages 229- direct income transfers to households which use the money
231). There is a slight difference between their procedure to increase their consumption. The total amount of the
and using the inverse of the CCF because the weights for transfer is determined so that the demand for foreign
each commodity accounting ratio are based on their exchange generated by the increase in consumption
average/marginal shares in trade in the UNIDO analysis expenditure is equal to the original increase in the supply
and on their average/marginal shares in consumption in of foreign exchange.
the L-M case. Expressed in this way, we see clearly that the SER is
In practice, the difference between the estimates may be simply the inverse of the CCF. However, this justification
negligible, but it is not clear a priori that this must be SO. of the UNIDO procedure does not satisfy all doubts,
It is not unusual for full sets of shadow prices to show because in certain respects it is not consistent with either
wide variations between individual consumption items or the details of their method or the fundamental ideas which
groups of commodities - notably between foodstuffs, are stressed elsewhere in the book.
imported manufactured goods, and non-traded goods and First, any increase in aggregate consumption must
services. involve an increase in demand for domestically produced
The role of non-traded items is ambiguous in this or non-tradable goods, such as housing, personal services,
version of the UNIDO approach, since intermediate transport and distribution. These in turn will require extra
inputs into the production of consumer goods are included capital inputs which may be imported. Thus, it is strictly
but it is difficult to deduce how one is supposed to treat incorrect to exclude all capital goods in following the
the complementary factor inputs. Thus, it is quite possible UNIDO procedure, though as a matter of practicality one
that the difference between the two values of the SER may might adopt the assumption that such imports would be
be significant, in which case it seems that the inverse of small relative to the imports of consumer and intermediate
the CCF will be more consistent with the remainder of the goods and can thus be safely neglected.
analysis than will the value derived from the UNIDO
Second, at the beginning of this discussion of the
procedure.
UNIDO approach, there was a deliberate ambiguity about
whether the weights used in computing the SER should
reflect average or marginal patterns of expenditure. Partly
The UNIDO approach this is an issue of implementation because it is difficult to
One assumption specified in the UNIDO derivation of the obtain reliable estimates of marginal expenditure shares
SER is that there should be no ‘reverse import whereas average shares present no such problems.
substitution’ except with respect to the capital goods. In However, there is also the standard issue of whether
effect this means that all extra foreign exchange is directly shadow prices should reflect the net impact of a marginal
spent on net additions to the supplies of non-capital goods change in the demand for or availability of a good or
in the economy. This seems a rather sweeping assumption, resource or whether they should be valid for assessing the

I08 Project Appraisal June 1986


Conversion factors and shadow exchange rates
impact of non-marginal changes. For most items a
marginal evaluation is appropriate, but the best choice is
less obvious in the case of the principal macro parameters The UNIDO approach seeks to
such as the SER, the shadow price of investment and the distinguish between the effects of a
basic shadow wage rate. project on aggregate consumption and
In the end, this decision must depend on the manner in
which these macro parameters will be used: whether for
distribution: this can only be
the evaluation of independent, relatively small projects or sustained in assessing secondary
in a broader assessment of sectoral allocation and effects of projects by ignoring
alternative development strategies. The UNIDO distributional considerations
procedure relies upon weights based on the composition of
marginal expenditures of foreign exchange.
Regrettably the case studies at the end of the UNIDO then be equal to the value of foreign exchange as
volume give no hint of how they expect the SER to be measured by the increase in aggregate consumption,
calculated in practice. Nonetheless, in Chapter 17 they resulting from the various methods of adjustment. Of
discuss the derivation of another national parameter, the course, an extreme version of this approach would assume
shadow price of investment, from information related to that the government will restrict itself to one method of
the macro-economic planning objectives of the adjustment, in which case the SER may be directly
government. estimated by tracing the consequences of this policy.
This, combined with the overall emphasis on aggregate
consumption as the goal of development planning, suggests Fourth, until now we have ignored distributional
that it might be more appropriate to use average considerations because, as far as possible, the UNIDO
expenditure shares as weights in constructing the SER approach seeks to distinguish between the effects of a
rather than marginal shares as specified in Chapter 16. project on aggregate consumption and on its distribution.
Third, in various sections of the book, the UNIDO However, this distinction can only be sustained in
authors stress that their approach is not based on a assessing the secondary or indirect effects of projects by
presumption of optimal government policies but on the ignoring distributional considerations. There are two ways
assessment of the effects of current policies which reflect in which the recognition of distributional issues may alter
the constraints and pressures operating on policymakers. the UNIDO procedure for estimating the SER:
However, the justification for their method of estimating Foreign exchange committed or uncommitted. So far, we
the SER which was outlined above relies upon an implicit have implicitly assumed that aggregate consumption and
assumption of optimality, unless we believe that the public income are equally valuable, which, within the
government will actually respond to the availability of L-M approach, is equivalent to assuming that the
extra foreign exchange in the manner described. Ignoring accounting rate of interest (ARI) and the consumption rate
distributional considerations, the assumption of optimal of interest (CRI) are equal or that the critical consumption
policies implies that, at the margin, the alternative uses of level is equal to the average consumption level. This in
foreign exchange should be equally valuable so that turn implies that there is no premium on savings or
estimation of the SER by reference to the increase in investment, so that the shadow price of investment must
aggregate consumption allowed by the extra foreign be I .o. Clearly, the implications of this assumption are not
exchange is perfectly consistent. Once we abandon this compatible with other aspects of the UNIDO analysis, at
assumption, we are obliged to choose between two least while we treat extra foreign exchange as being
approaches. uncommitted in the sense that the government is able to
Maximum social value. We can define the SER as a use the revenue in whichever manner will generate the
measure of the maximum social value of one unit of highest social return.
foreign exchange subject to certain constraints on the Thus, we have to choose between either the L-M view
alternative policies. In this case, the value of the SER as of foreign exchange as uncommitted public income which
calculated in the manner described should set a lower will be used to increase investment rather than aggregate
bound on the ‘true’ SER, provided that the income consumption so long as the shadow price of investment
transfers envisaged are regarded as being feasible. exceeds 1 . 0 , or a strict interpretation of the suggestion in
the UNIDO manual that changes in the availability of
Deterministic view. Alternatively, we can adopt a more foreign exchange will have no influence on the amount or
deterministic view of policy and estimate the SER by composition of investment.
assuming that the government can utilise additional foreign In the latter case, extra foreign exchange cannot be
exchange in only a small number of ways - such as by regarded as being ‘uncommitted’ in the L-M sense, so
increasing public investment, lowering taxes, increasing there need be no simple relationship between the SER and
public employment or the minimum wage level. In this the CCF though for the special case in which foreign
case, the policy option of direct income transfers which exchange is committed to aggregate consumption via a
will increase aggregate consumption must probably be proportional income transfer, the equivalence between the
ruled out and instead we must examine the net impact of SER and the inverse of the CCF remains valid. Under the
each method of adjustment including any effects on L-M assumptions of uncommitted foreign exchange, it is
investment and imports of capital goods. The SER should straightforward to show that:

Project Appraisal June 1986 I09


Conversion factors and shadow exchanae rates

SER = @ /CCF This brings us back to the point stressed above, that is,
that we are concerned with uncommitted public income so
that income which is spent in ways outside the control of
the government must be analysed more specifically to
evaluate the net impact of the expenditures. Thus, it is
where wc, wa are the income weights for the critical neither helphl nor appropriate to use a separate shadow
consumption level and for average consumption, and ICF exchange rate adjusted to allow for the non-optimal level
is the typical conversion factor for investment. of foreign borrowing. Instead, the standard shadow
Constraints on foreign exchange use. When the manner in exchange rate - the inverse of the CCF - should be
which extra foreign exchange can be used is constrained, used in combination with separate valuations of the social
we must typically allow for the specific distributional costs of the constraints which are the cause of the over-
consequences of the alternative allocations considered in dependence on foreign capital.
estimating the SER. For example, if the adjustment to an
increase in the availability of foreign exchange proceeds Standard conversion factors
via an increase in employment and wage payments, then
the distributional weights appropriate to the beneficiaries Just as those using shadow exchange rates have too often
of this policy should be used in computing the SER. been somewhat careless about the proper specification of
The alternative approach would require that we attempt their numeraire, so also have practitioners within the
to combine the direct distributional effects of a project Little-Mirrlees school tended to be rather inexplicit about
with the indirect distributional effects associated. with the the scope of their standard conversion factors. This is
adjustment to the consequential change in foreign important because it is widely believed that the shadow
exchange availability or with other indirect effects of the exchange rate should simply be the inverse of the standard
project. This procedure is unlikely to be practicable even conversion factor - see, for example, Pursell’s
when applied to only moderately complex projects. It is comparison of his shadow exchange rate for the
precisely because of the difficulty of carrying such Ivory Coast with Linn’s SCF (Pursell 1978, pages 43-
decompositions that aggregate shadow prices have been 50). In most cases, this view is incorrect, because it

adopted as an efficient means of valuing the wider effects involves confusions both about the numeraire and about
of changing the supply/demand for important resources or the role of the SCF, as the comparison of the UNIDO
inputs. and L-M analyses in the previous section has already
shown.
Excessive foreign borrowing. The UNIDO manual’s in the estimation of specific accounting prices. Instead of
second method of deriving the SER is based on the trying to estimate accounting prices for all items involved
treatment of foreign exchange as a merit item in the in decomposing the social codbenefit of some
government’s objective function. It is suggested, for commodity, it is suggested that an average ratio of
example, that extra foreign exchange revenue reduces accounting prices to market prices could be used. L-M
dependence on foreign loans or graQts (public and private) suggest (1974, page 218) that in principle the average
and thus generates a social benefit in addition to the direct should be derived from a calculation of the form
contribution of the foreign exchange to aggregate
consumption.
This implies that foreign borrowing is above the level
SCF =
CQ i. AP
regarded as desirable by the government, so that it is CQ i. M P i (3)
willing to sacrifice aggregate consumption to reduce capital
inflows. Such a view of dependence on foreign capital is
where
quite common in developing countries, but it is not clear
that a higher shadow exchange rate is the appropriate
APi = accounting price of good i,
MP; = market price of i ,
method of allowing for the underlying reasons for this
dependence. Qi = total supply of i .
It must be associated with institutional or political This is clearly likely to be too laborious in practice, so
factors which limit the government’s scope for that, following Scott’s work on Kenya (Scott 1974, page
intervention, since otherwise it could use direct taxation to I 711, they suggest that the median or the modal value of
reduce aggregate consumption and to free resources which the distribution of accounting ratios could be used as the
could be used to improve the balance of payments. SCF. Both L-M and Scott then comment that the inverse
However, the existence of such constraints may not be of such a SCF would be equivalent to a shadow exchange
consistent with the initial assumption that extra foreign rate estimated with respect to the value of some aggregate
exchange can be used to reduce foreign borrowing. This basket of goods at market prices as the numeraire.
may, of course, be the case, but it seems probable that the Provided that the SCF, the SER, and the numeraire are
outcome will depend upon the manner in which the carefully and consistently defined, this may indeed be true,
foreign exchange is earnedand the recipients of the income but it is not very helpful because the difficulty really lies in
associated with the activity. In other words, we must the achievement of the level of detailed analysis and
examine the ways in which the revenue and incomes are consistency required.
committed to various categories of consumption and other Scott’s estimates for Kenya and my own work in
expenditure. various countries indicate that for non-traded goods and

II0 Project Appraisal June 1986


Conversion factors and shadow exchange rates
services and, to a lesser extent, for imported manufactured
goods, the accounting ratios tend to cluster reasonably Experience suggests that traded
closely about the median value. Unfortunately, this tends agricultural and primary commodities
not to be the case for exportables and agricultural usually have accounting ratios which
products. As a result, choosing a ‘typical’ accounting ratio are very different from those
to encompass the whole range of accounting ratios may be
seriously misleading, while the calculation of an average estimated for traded manufactured
valueas in equation (3) involves serious conceptual goods as a result of various forms of
difficulties. government intervention
For example, it is not clear whether QIshould be
defined as total domestic availability - production plus expenditures on non-traded goods and services. Again one
imports - or total domestic consumption - production can use the median value of the distribution of accounting
plus imports minus exports. Presumably, we should use ratios for non-traded goods and services. Alternatively, an
total supply rather than final demand but this implies that average conversion factor may be estimated by aggregating
the resulting SCF may depend upon the degree of input-output data for all sectors which predominantly
disaggregation of inter-industry trade in the data available. supply non-traded items and then performing the
Further, the set of items to be included in the average is calculation in the usual way.
not immediately obvious. Should it include commodities
It is necessary to bear a number of points in mind when
which are exclusively used as inputs into the manufacture
considering this suggestion or using standard conversion
of traded items? These issues can, of course, be resolved
factors.
once the purpose of estimating a SCF is decided, but up
until now insufficient thought has been given to them. First, the procedure for estimating the TSCF
Squire and van der Tak (S-T) introduce their SCF for specifically focuses on manufactured goods. This is
use in converting minor direct and indirect non-traded because experience suggests that traded agricultural and
inputs to accounting prices when estimating the accounting primary commodities usually have accounting ratios which
price of an item by decomposition. In addition, they are very different from those estimated for traded
suggest that the SCF may be approximated by taking “the manufactured goods as a result of export taxes, low tariffs
ratio of the value at border prices of all imports and on imports, consumption subsidies, and other forms of
exports to their value at domestic prices”. government intervention. Not only do these items ’have
Unfortunately, this merely serves to compound the untypical accounting ratios as a group, but also there tends
confusion concerning the use of standard conversion to be a high degree of variation between the accounting
factors, since it assumes that the accounting ratios for ratios for the commodities making up the group.
non-traded goods may be approximated by the accounting Further, in most cases, the principal items for which
ratios for traded goods. the TSCF would be used are miscellaneous semi-
It is possible to specify a set of assumptions which manufactured or manufactured goods which are required
would justify this approach, but they are quite implausible as inputs in the production of goods and services. If they
and imply that the accounting ratios for the unskilled form a significant proportion of costs, primary commodity
labour and profit components of domestic value. added inputs should be converted to accounting prices by the use
should be little different from the typical accounting ratio of specific accounting ratios. Minor expenditures on
for traded goods. This greatly narrows the scope for primary commodities may be treated either as purchases of
studying the implications of using accounting ratios for non-traded agricultural output, or, if the other methods
particular domestic resources which differ substantially would be inappropriate, by applying the TSCF.
from those for traded goods. Second, a common feature of tariff structures in LDCs
as well as in developed countries is the escalation of the
Three general conversion factors relevant tariff as the stage of manufacture of the imported
item increases. In other words, raw materials like raw
In order to clarify the role of standard conversion factors,
cotton and wool, coal, copper and other mineral ores all
it seems advisable to work in terms of at least three distinct
bear low tariff rates, whereas higher rates are imposed on
general conversion factors:
semi-manufactured and intermediate goods - inorganic
The Consumption Conversion Factor (CCF) defined and and organic chemicals, aluminium billets, steel coil,
used in the manner specified by both L-M and S-T. electric motors, vehicle components - and still higher
rates on finished manufactured goods.
The Standard Conversion Factor for Traded Goods
The extent of such escalation depends upon the extent
(TSCF) which would be used to convert minor items of
to which the government is trying to encourage import
expenditure on importable commodities to accounting
substitution via the tariff structure. If the tariff rates differ
prices. Its value may be calculated either by using the
significantly between these categories, then it may be
median tariff on imported manufactured goods in the
sensible to use the TSCF for finished manufactured goods
standard procedure for estimating the accounting ratio of
subject to the relatively higher tariff rates and to estimate a
an importable or by taking the median accounting ratio for
separate general accounting ratio for intermediate and
non-agricultural imported commodities.
semi-manufactured goods.
The Standard Conversion Factor for Non-traded Goods Third, the non-traded items for which the NSCF is
(NSCF) which would be applied to miscellaneous used are likely to be dominated by various types of

Project Appraisal June 1986 111


Conversion factors and shadow exchange rates
services. Assuming that distribution, transport, electricity,
construction and similar major non-traded inputs are The Little-Mirrlees-Scott method is
treated separately, it is probable that financial services - relatively straightforward while at the
banking, insurance - and miscellaneous administrative same time taking account of many of
overheads for projects will be important components of
the remaining non-traded expenditures.
the criticisms which have been
Unfortunately, these are conceptually difficult to deal levelled against the use of conversion
with and the data relating to them - whether from input- factors
output tables or other sources - is usually scarce and of
doubtful reliability. The payments for such services will Thus, in the remainder of this paper, I shall assume
also include substantial transfer payments or expenditures that shadow exchange rates are used to convert foreign
on the replacement of insured property and goods which exchange expenditures into their social value in terms of
are not treated as part of the output of the relevant service aggregate consumption. In the general case when we are
industries in constructing input-output tables or national moving from the L-M numeraire of uncommitted public
income accounts. income measured in terms of foreign exchange to
In addition, much of the value-added in these industries aggregate consumption numeraire, this implies that the
is contributed by skilled labour, whose accounting ratio is standard equivalence between accounting ratios and
probably the most difficult to estimate of all of the shadow exchange rates is:
accounting ratios for the major primary inputs. All of
SER = (Ratio of the income weight at the critical
these considerations indicate that the NSCF may have to
consumption level to the average income
be estimated in a rather rough and ready manner.
weight reflecting the distribution of
For the financial and other non-traded services - after
aggregate consumption) / C C F
allowing for transport payments, and so on - I would
suggest the use of a weighted average of the accounting If, however, the government is viewed as having only
ratios for skilled and unskilled labour, construction and limited ability to control the utilisation of extra foreign
the TSCF. The weights would depend on the analyst’s exchange, the income weight at the critical consumption
judgement concerning the relative importances of the level in the formulation above must be replaced by an
various items in net output. aggregate income weight reflecting the best of the
In previous studies, I have combined these items in a alternative methods of allocating foreign exchange.
category ‘Business Services’ (Hughes 1979, I 980) rather
than including them in the NSCF, but the estimation
procedure followed was that described above. Other non- The Little-Mirrlees-Scott method
traded goods and services can be analysed in the usual way
and a composite accounting ratio estimated for use as the As experience with the application of the Little-Mirrlees
NSCF, either by taking the median valueor some method of project appraisal has built up, procedures for
appropriately weighted average. implementing it have been refined and in this section I will
The fourth point in using standard conversion factors briefly describe the current state of the art. The fullest
follows immediately from the previous discussion. None of application of these procedures has been in Kenya for
the standard conversion factors specified here can be used which some 200 accounting ratios have been estimated
to derive an implicit shadow exchange rate with respect to (Scott, MacArthur and Newbery, 1976; Little and Scott,
aggregate domestic expenditure at market prices as 1976).
numeraire. In any case, for similar reasons to those Scott’s description of the way in which the basic set of
outlined in discussing the formula for the SCF in (3), I do accounting ratios for Kenya was estimated is the best
not think that this is a very helpful numeraire by practical guide available (Scott 1976). His methods are
comparison with either the one used here or the UNIDO relatively straightforward while at the same time they take
one. account of many of the criticisms which have been levelled
This brings us back again 10 the question of the against the use of conversion factors, such as failure to
numeraire when a shadow exchange rate is used. allow for demand or supply elasticities, allowances for
Normally, the shadow exchange rate is defined simply as externalities or economies of scale. For this reason I will
the social benefidcost to the economy of refer to the current approach as the Little-Mirrlees-Scott
earning/spending an extra unit of foreign exchange. For (LMS) method.
Bacha and Taylor (1971) the ideal shadow exchange rate In estimating a set of accounting ratios, we start by
would come from a comprehensive programming model, distinguishing between a category of primary inputs and all
from which one may infer that they have current aggregate other goods and services. Primary inputs are those items
consumption in mind as their numeraire, since this is the whose accounting prices are determined either by
typical unit of welfare measurement in such models. considerations which cannot be incorporated in the
Similarly, at several points in his article on shadow simultaneous equation framework to be outlined - such
exchange rates, Balassa (1974) indirectly refers to the as the treatment of foreign exchange or certain categories
value of foreign exchange in terms of consumption; for of saving as merit items - or by the application of
instance “the shadow exchange rate will be employed to relatively complex formulae/models involving assumptions
convert . . . the consumption cost of factor use into world that one might wish to test in sensitivity analyses - such
prices”. as various categories of labour, profits.

I12 Project Appraisal June 1986


Conversion factors and shadow exchange rates
The accounting ratios for these primary inputs are benefit analysis. If this were not so, then the elements of
determined with the aid of exogenous data and are then the matrices A and B would not be invariant to the scale
used to derive the accounting ratios for other items within of demand.
a quasi-input-output framework. Applying the familiar Note, however, that this is not the same as assuming
L-M principles, the market prices of these other items are constant average costs, so long as it is recognized that the
decomposed into expenditures on primary inputs and accounting ratios are not appropriate when large changes
other goods or services. Thus, we can construct a pair of in the supply or demand for a particular item are
input matrices in value terms: envisaged. In such cases, as always in project evaluation,,
specific analyses must be carried out for those items whose
social cost might be significantly affected by large scale
A = [a i,l changes.
shows the market values of goods and services entering as Otherwise, the assumption implies constant marginal
direct inputs into the social codbenefit of costs, so that the economies or diseconomies of scale may
supplying/consuming one rupee's worth of each item. be incorporated by using input coefficients a,, bth which
Thus aij is the cost at market prices of good j as a differfrom the average input requirements.' This is still
component in the social cost of one rupee's worth of good restrictive since it rules out the possibility that marginal
i at market prices. costs may fall or rise continuously with the scale of
production.
Probably such a pattern is fairly unusual, but when it
shows the market values of primary inputs entering as seems important enough to worry about it may be dealt
direct inputs into the social cost of using one rupee's with by distinguishing between several accounting ratios
worth of each item. In this case the subscript h. denotes for the commodity, each of which is estimated on the basis
the primary inputs. of the commodity's cost composition over some range of
Now suppose that the primary input accounting ratios total output. In effect, this means that one would follow
are denoted by W h (h = I , . . . m) and the commodity the usual method of linear approximations used in the
accounting ratios by vj 0' = I, . . . n). Then, we have the programming literature.
equation :
Inelastic world supply or demand
v = Av + Bw (4)
Cases in which a country faces demand or supply curves
in which v, w are the vectors of commodity and primary which are not perfectly elastic can similarly be allowed for
input accounting ratios respectively. This yields: within this approach by adjusting the input coefficients.
v = [I -A]-' Bw Suppose that the country exports cotton yarn at an average
cost, insurance, freight (cif) price of Rs p r per unit,2 but
or (5) faces a foreign demand curve having an elasticity of price
v = Rw, where R = [i-AI-' B with respect to exports of - E ( E >o). Thus, if exports are
increased, the net foreign exchange earnings per unit are
with R being the matrix of total direct and indirect
pp( I - E ) while existing exporters lose ~ p in, terms of lower
requirements of the primary inputs.
profits or lower producer incomes.
It follows that if we are given or can estimate values for
Suppose the domestic ex-factory price of cotton yarn is
the primary input accounting ratios, then the remaining
p d and that the transport and other costs of exporting the
accounting ratios can be obtained very simply once the A
yarn are ce. So, we can write the social cost of the cotton
and B matrices have been constructed. In fact, some of
yarn sold in the domestic market as:
the primary input accounting ratios themselves depend on
the commodity accounting ratios, so that an iterative v$d = (I-€) pewf + Epewp + cpt (6)
procedure is followed.
where
We start by assigning reasonable, though arbitrary,
v , = accounting ratio (AR) for cotton yarn
values to the primary input accounting ratios and use these
wf = AR for foreign exchange (this will be 1 . 0 unless
to obtain commodity accounting ratios. These are fed into
foreign exchange is viewed as a merit item)
the estimation of the primary input accounting ratios, and
w p = AR applicable to the increased profits or
the whole procedure is repeated until successive estimates
incomes due to the reduction in exports
of the primary input accounting ratios converge.
vt = AR for transport and miscellaneous costs.
Experience suggests that two or three iterations are
sufficient to estimate these accounting ratios to a Dividing (6) through by the domestic price p d gives the
reasonable degree of accuracy. input coefficients for and A and B matrices.
It is important to recognize the assumptions underlying Two points about this procedure should be stressed.
this approach and also the range of factors affecting social First, we see that it allows for the imperfections of world
codbenefit which may be incorporated in the analysis. demand or supply facing the country and thus invalidates
the common suggestion that procedures based on
Constant costs versus economies/diseconomies of scale
conversion factors assume infinitely elastic world supply or
The procedure assumes that the separate items making up demand.
the social costs of a good or service vary linearly with the Second, in adjusting for market imperfections, it
amount of the goods involved for the purposes of cost- prompts us to take proper account of any distributional

Project Appraisal June 1986


Conversion factors and shadow exchange rates
changes brought about by the price changes. This is the In the case of some projects or countries there may be
role of the element €pewpin (6) which may in some cases real uncertainty about whether a good should be treated as
be quite significant and which is often neglected in other an importable or an exportable or the pattern of trade may
analyses. be expected to switch in the medium term. For specific
Thus, we may observe that the LMS procedure has the projects, it is not difficult to allow for this by calculating
merit of focusing our attention on the indirect effects of and using two accounting ratios appropriate to the two
producing or using certain commodities which might patterns of trade. However, for a general set of shadow
otherwise be neglected. Indeed, it is possible to take the prices, the real problems arise if the commodity is the
analysis of both this point and of economies of scale one major foodstuff in domestic consumption and there is a
step further by allowing for the possibility that changes in large difference between the import and export parity
the supply of or demand for such goods may change prices. This was the case with maize in Kenya for a
domestic prices and thus indirectly lead to changes in non- number of reasons associated with agricultural change,
project domestic consumption and/or production whose government policy and investment in transport facilities.
net social value should be incorporated in the shadow The appropriate procedure here is to assume a market
price (Scott 1976, page 194). price for the commodity between the import and export
Inelastic domestic supply parity prices and then to calculate two accounting ratios
for it on the alternative assumptions about the direction of
Items subject to elastic domestic supply may be divided trade. These imply two different sets of accounting ratios
into agricultural and other product, remembering, of which define the limits of the variation in shadow prices
course, that such constraints are only relevant so long as due to this problem. the choice of which set of accounting
the item is non-traded. For non-traded agricultural ratios to use must be based on a judgement about the
output, the usual LMS procedure is to treat this as a likely pattern of trade over the next 5-8 years.
general primary input, whose accounting ratio is estimated Alternatively, an accounting ratio for the foodstuff in
by special analysis. between the two extreme values may be adopted in order
Scott (1976, pages 156-166) discusses the factors to derive a single set of accounting ratios.
which affect the social cost of non-traded agricultural
I have discussed the treatment of these various features
output and gives a rather complicated formula which takes
of trade and production at some length in order to
account of substitution between traded and non-traded
illustrate the point that the LMS method of estimating
goods in agricultural production, effects on the distribution
shadow prices encourages the analyst to examine and allow
of income, substitution in consumption patterns, and the
for all the various institutional and other factors which
costs of marketing. However the accounting ratio for non-
affect the relationship between social and market values. It
traded agricultural output is estimated, it is bound to be
is often suggested by advocates of shadow exchange rates
subject to a considerable margin of error because of
that procedures which focus on shadow prices for goods
difficulties with data as well as the analytical complexity of
fail to allow for variable returns to scale, imperfect world
the factors involved.
markets, etc. Quite clearly, this argument is incorrect.
Nonetheless, by distinguishing. the category as a
Indeed, it may be turned on its head, because the
separate primary input we can easily examine how
procedures outlined enable us to incorporate the specific
sensitive the other accounting ratios are to variations in our
effects of these factors for particular commodities, whereas
assumptions about the social costs of non-traded
they can at best be allowed for via estimates of shadow
agricultural commodities. If it turns out to be important,
exchange rates in a crude and aggregative manner.
then more effort can be devoted to improving the estimate
This leaves the possibility that there are certain general
of the accounting ratio without disrupting the other work.
factors at an aggregate level which are not adequately
Non-agricultural products subject to inelastic supply
encompassed by the micro-level analysis of the LMS
may be treated in the same way as either non-traded
procedure. I have discussed tlis in another paper, but note
agricultural output or products subject to increasing costs
that this would imply a real problem of consistency within
of production. In my experience, the only item which
a framework based on shadow exchange rates as
warrants specific treatment similar to non-traded
conventionally specified. These are, in effect, estimated as
agricultural output is the social costs of urban house rents,
an index of the ratio of domestic-to-border prices, using
since an increase in the demand for urban housing can
marginal trade flows as weights - eg Balassa (1974,
generate large distributional effects which should be
equation (7a) ). If, in addition to the use of this conversion
incorporated in the analysis. Other non-traded goods or
factor for foreign exchange, specific adjustments are made
services with inelastic supply may be dealt with by
to allow for inelastic world demand/supply in the cases of
combining the treatment of diseconomies of scale with the
particular commodities, then clearly there is an element of
analysis of price changes in the previous two sections.
double counting in the final shadow prices of those
Switches in trade jlows commodities.
In general, an economist estimating a set of shadow prices On the other hand, it would be even more inaccurate to
must make a judgement about whether particular goods fail to adjust for the specific market imperfections and it
should be treated as exportables, importables or non- would be impossibly complicated to attempt to eliminate
traded goods. This judgement must be based on an the double counting
analysis of the medium term prospects given the policies The point may be put in another way. Shadow
taken as fixed in our discussion of the second best above. exchange rates were originally derived as a short-cut

1x4 Project Appraisal June 1986


Conversion factors and shadow exchange rates
method of allowing for the divergence between social Social Costs
values at border prices and at domestic prices. However,
they are not designed to allow for such divergences at a First, shadow prices should reflect the social costs or
values of items to the economy on a medium or long term
micro-level, whereas the Little-Mirrlees method starts
view. This implies that we should incorporate the effect of
from that level and builds upwards.
the devaluation on the prices of domestically produced
Hence, as we shall see, it is not normally necessary to
goods and services in calculating shadow prices. Hence,
use an aggregate shadow exchange rate in a procedure
we would need to follow through the full ramifications of
based on properly estimated accounting ratios. The
the higher prices for traded goods and services, not only
UNIDO approach, as shown above, needs a shadow
with respect to the behaviour of domestic producers faced
exchange rate only because of its use of a different
with an increase in the nominal cost of intermediate inputs
numeraire, so that its shadow exchange rate is simply the
but also in terms of the adjustment of wage rates, profit
inverse of the accounting ratio for its numeraire in. the
margins and other factor incomes.
LMS approach.
If producers set their prices according to some kind of
markup principle and factor incomes in real terms are
unaffected by the devaluation, it can be shown that the
Exchange rate changes LMS method of calculating accounting ratios may easily
be modified to yield estimates of the post-devaluation
The discussion so far has been based on the assumption shadow prices.
that the current money exchange rate corresponds to the In order to demonstrate this point, it is first necessary
long term equilibrium exchange rate given the system of to distinguish between two sets of accounting ratios (ARs)
taxes, tariffs and subsidies taken into consideration when which will be referred to in the discussion.
calculating the shadow prices. The concept' of an
Current price ARs, denoted by v;, wh, give the ratios of
equilibrium exchange rate is not well defined for
the shadow prices of the items at some future date to their
developing countries.
market prices at that date;
Here I shall assume that, given world prices and market
prospects, the exchange rate is at its long term equilibrium Constant price ARs, denoted by v*;, w*h, give the ratios
if, when the level of economic activity in the economy is of the shadow prices of the items at some future date to
such as to ensure full capacity utilisation in the long term their market price at date o of the analysis.
for the modern sector, the balance of payment shows a
Obviously, we need to know the current price accounting
deficit (surplus) which corresponds to the government's
ratios if the work of estimating shadow prices is to be
willingness to borrow abroad (to acquire foreign assets).
useful for the purpose of evaluating proposed projects or
In practice, those concerned with the estimation of
policies in the future, but it is also convenient to calculate
shadow prices may feel that the current exchange rate
constant price accounting ratios since these provide us with
diverges from the equilibrium rate in this sense, even if the
information about the magnitude of the adjustments
balance of payments is not systematically in surplus or
associated with a move to an equilibrium exchange rate.
deficit. For instance, the balance of trade may be
Let us suppose that the percentage devaluation required
maintained at a certain level by the use of temporary taxes
to move from the current to the equilibrium exchange rate
and tariffs, or by quantitive controls.
is I ooy , Then, the LMS method will yield the current and
Temporary measures should clearly be excluded from
the constant price accounting ratios if we adopt an
the estimation of shadow prices while we have assumed
appropriate normalisation of the accounting ratio for
earlier that quantitive restrictions should not be taken into
foreign exchange:
account when constructing our estimates. Alternatively, it
may be clear that the country cannot continue in the long For convenience, define a trade-weighted basket of
term to borrow abroad at the current rate or that the foreign currencies as the unit for border prices in the
exchange rate can only be sustained by restricting the level analysis - denoted by the abbreviation BPU; and
of economic activity considerably. define e, as the number of units of domestic currency
For whatever reason we may therefore have to take purchased by 1.0 BPU at date t .
account of the divergence between the current exchange Suppose that the project for which er is at its maximum
rate and the equilibrium rate. It will be assumed that the is implemented at date t = s . Denoting the accounting
current exchange rate is overvalued and that the amount of ratio for foreign exchange at date t by w,(t>,we choose
the devaluation required to eliminate the divergence has the normalisation:
been separately estimated. These are important issues, of w,(s) = 1.0
course, but they are not specifically related to our For any other projects at date t the accounting ratio for
consideration of the methodology of project evaluation. foreign exchange should be given by:
The usual recommendaton then is that the shadow
prices should be calculated by adopting the equilibrium w,(t> = er/et
exchange rate in place of the current exchange in This procedure ensures that all projects are converted to
converting all foreign exchange rate as illustrated by the border price terms on an equivalent basis. It is then
discussion in Bacha and Taylor (1971).This is, however, only necessary to deflate the net present value of
not a satisfactory method of adjustment for two different projects by an index of commodity prices
interrelated reasons. measured in border price units.

Project Appraisal June 1986 115


Conversion factors and shadow exchange rates

be achieved either via a general increase in taxation or


If the devaluation leaves relative by holding- down the nominal wage - rates paid to
prices and factor incomes unaltered some/all workers while allowing prices to respond to
there may be little point in adjusting the devaluation.
accounting ratios because the Whatever the mode of adjustment, it will alter the
adjustment will have no effect on the accounting ratios for wages and other factor payments by
ranking of alternative projects changing the magnitude and the composition of the
resources committed as a result of offering additional
employment. The consumption conversion factor would
For example, suppose that we choose the US$ as our
be reduced by an increase in indirect taxation, which
BPU for a country and that at present e, = 5 . 0 whereas would in turn reduce the income weights for specific
the medium term equilibrium exchange rate is estimated to
income levels. The overall switch of resources from
be e, = 6.0. Then, for all projects evaluated at pre-
personal consumption to investment and public
devaluation market prices, we should use the constant
expenditure implied by a reduction in real wages means
price set of accounting ratios which should be calculated
that in this case also the ARI will increase relative to the
by using an accounting ratio for foreign exchange of wj(01 CRI.
= I .2. The current price set of accounting ratios is
We have seen that the elimination of a long term
required for the evaluation of projects whose financial
balance of payments deficit requires adjustments in real
information is based on post-devaluation market prices
income and expenditure levels that would affect the long
and for this purpose the accounting ratio for foreign
term values of key national parameters in a set of shadow
exchange will be wj(s> = 1 . 0 . prices, which would in turn also change most of the
It is not sflicient to modify only the accounting ratio
accounting ratios fdr goods and services. Hence, an
for foreign exchange in calculating the constant price set of
economist charged with the task of constructing a set of
accounting ratios, since this change will also affect the
shadow prices for a country which appears to be suffering
accounting ratios for the other primary inputs as well as
from a fundamental balance of payments disequilibrium
the accounting ratios for goods and services.
must start by considering the alternative modes of
In fact, it is straightforward to show that, if the
adjustment which might be adopted for the purpose of
devaluation leaves relative prices and factor incomes
achieving long run equilibrium.
unaltered, all of the constant price accounting ratios will
It may then be necessary to construct several alternative
simply be equal to the accounting ratios which would be
sets of long run accounting ratios - current price
obtained if the exchange rate adjustment had been ignored
accounting ratios - corresponding to different
multiplied by the ratio eJe, as described above. This
combinations of adjustment policies. In practice, the
implies, of course, that there may be little point in
effects of the alternative policies on the important national
adjusting accounting ratios for a devaluation of this kind,
parameters are likely to be rather similar, so that one or at
because the adjustment will have no effect on the ranking
most two sets of accounting ratios will be sufficient to
of alternative projects.
represent the expected long term pattern of relative social
The discussion above confirms the results obtained
costs and benefits of the items under consideration.
from current macro-models of exchange rate adjustment.
A devaluation on its own is not sufficient to eliminate the Cases where adjustments are inappropriate
macro-economic disequilibrium which is the source of the
There are, however, circumstances in which it would not
original balance of payments deficit. To be effective, the
be appropriate to incorporate long term adjustments of the
devaluation must bring about a change in the relative
type envisaged above in the accounting ratios computed to
prices of traded and non-traded goods and reduce
assist in the appraisal of projects over a period of years.
aggregate income and expenditure levels at full
One obvious case arises when a country is operating a
employment.
conscious policy of borrowing abroad, relying upon short
Hence the second reason for the unsatisfactory nature
or long term capital flows to offset a trade deficit.
of the earlier method of adjustment is that, in constructing
Equally, a country which is experiencing a temporary
a set of accounting ratios which take account of the need
balance of payments deficit because of lags in the process
to eliminate initial disequilibrium, we must focus on the
in adjusting the exchange rate in response to a domestic
real changes in the pattern of incomes and expenditures
rate of inflation in excess of the level of world inflation as
which are associated with the adjustment. There are, in
measured by the foreign exchange costs of traded goods,
practice, two methods of adjustment:
should not be treated as a country with a fundamental
0 Reduce investment and/or public expenditure. This balance of payments disequilibrium.
implies that the value of public income relative to In such a case, it is nonetheless necessary to make some
aggregate consumption must rise, so that income allowance for the prospective exchange rate change. This
weights must fall, and also the accounting rate of should take the form of estimating shadow prices on the
interest (ARI) will increase relative to the consumption basis of a ‘normal’ set of market prices, that is, a set of
rate of interest (CRI), since the gap between the critical prices which reflects the pattern of relative prices which
consumption level and the average consumption level would prevail if the exchange rate was indexed to the
has increased. difference between domestic and world inflation rates.
0 Reduce real wages and other factor incomes. This may Perhaps the most interesting case in which it may not

I 16 Project Appraisal June 1986


Conversion factors and shadow exchange raies

be necessary to take account of the real changes production - will be increased by the changing
accompanying an exchange rate change arises when we composition of production. Alternatively, it may be
consider the role of exchange rate policy as an instrument feared that serious shortages of key categories of skilled
of macro-economic control. labour will develop.
The discussion above was predicated on the assumption
We will return to both these factors in due course.
that a devaluation which has no long term real effect on
Excluding their influence, extra production or investment
relative prices and factor rewards cannot resolve the
which satisfied our criterion of social profitability on the
original disequilibrium which prompted it. If we assume a
basis of long run shadow prices must improve the balance
given structure of protection and a constant level of
of payments.
capacity utilisation or unemployment, this is indeed
The argument above shows that even if long run
correct. However, neither of these assumptions may be
relative prices and incomes are not altered the devaluation
appropriate for most developing countries, and even if
may fulfil its purpose. In effect, it provides a respite from
they were, the devaluation may be regarded as a valuable
the short term pressure of balance of payments difficulties,
component of short or medium term economic
while longer term policies of investment and structural
management.
change deal with the underlying disequilibrium. Even in
Effects of dev a1u ation the absence of investment, the re-establishment of the
original set of relative prices and incomes need not mean
It is not possible here to develop a full analysis of the
that production and employment decline back to their
relationship between exchange rate policy and investment
initial levels.
analysis, but I will sketch the main lines of the argument.
If the extra production is sufficiently profitable in terms
Suppose that the devaluation is envisaged because of the
of shadow prices, this alone may be enough to greatly
need to adjust to a current or prospective balance of
reduce or eliminate the disequilibrium. This, of course,
payments deficit or as an accompaniment to the
illustrates the fundamental point that we should not
dismantling of quantitative trade restrictions. Assuming
assume that the nature of long run
that the level of investment is determined separately then
equilibrium/disequilibrium is invariant to the policies and
the devaluation can only succeed by altering the balance
other factors influencing short run dynamics.
between factor productivity and factor incomes - or,
The analysis above applies only if the underlying
more accurately, consumption out of factor incomes.
disequilibrium is relatively small. It would take too long to
In the short run, a devaluation will typically reduce real
eliminate the fundamental balance of payments
wages and thus affect this balance directly. Further, its
disequilibria experienced by some countries by the kind of
effects on relative prices will enhance the profitability of
production and investment gains discussed. Further, while
producing tradable goods while increasing the demand for
the devaluation may leave real wages and consumption
non-traded goods. Thus, we may expect an increase in
unaffected in the long run, it will still cause changes which
capacity utilisation and employment throughout the
should be reflected in the shadow prices used for the
economy, assuming that this is possible. Provided that,
estimation of project costs/benefits in the short term.
when valued at shadow prices, this extra production is
There is a general problem concerning the appropriate
socially profitable, it will contribute to the country’s net
method of constructing and using shadow prices which
resources and thus improve the balance of payments.
vary over time. Too little attention has been given to this
Similarly, as the short term extends to the medium and
problem in the past, but I will leave a proper treatment of
long term, new investment projects will come to fruition.
it to another paper and will focus on the consequences of
Provided that these satisfy the criterion of social
the temporary effects of a devaluation.
profitability using long run shadow prices, they too will
For the project analyst the most significant short term
contribute more to the country’s aggregate income than
change caused by a devaluation may be assumed to be a
will be committed to consumption and other expenditure.
reduction in the real wages and consumption of non-
Further, as long as real wages remain below their long
agricultural workers which implies that shadow wages
term level, this contribution will be increased.
must fall relative to other shadow prices. It follows,
Against this, there are two factors which might reduce
therefore, that if we use the shadow prices calculated on
the balance of payments benefits of the increase in capacity
the assumption of no change in relative prices and
or production. These are:
incomes, we will, in most cases, tend to be too
The benefits which make the production or investment conservative about accepting projects.
socially profitable may be distributional rather than With very few exceptions, projects which are socially
resource benefits. In principle, this should not matter profitable on this basis will be at least as worthwhile when
provided that the government is prepared to increase
taxes on consumption or incomes to offset the
commitment of resources to consumption. In practice,
If we use the shadow prices calculated
of course, it may be very reluctant or unable to do this. on the assumption of no change in
As employment and capacity utilisation rise, the input relative prices and incomes we will
coefficients used in calculating the shadow prices may tend to be too conservative about
be significantly altered. In particular, it may be thought accepting projects: if the devaluation
that the marginal product of labour in agriculture - or, is taken into account they will be at
more generally, the social opportunity cost of labour in

Project Appraisal June 1986


Conversion factors and shadow exchange rates
evaluated at shadow prices revised to take account of the non-agriculture wages, @ .
effects of the devaluation. The exceptions will be those 0 On the assumption the accounting ratio for
projects whose primary benefit is derived from the direct marginal urban household expenditure is similar to
or indirect replacement of labour in production or the consumption conversion factor, vo calculate
consumption by capital or other inputs - such as labour- the change in the accounting ratio for urban wages
saving mechanisation or the displacement of pottery or using:
wooden consumption goods by plastic or metal-substitutes.
Given that even the short run real effects of a
Aw*,, = -$(I-") (I-uA
~ * c (8)
devaluation are often quite uncertain, it may be sensible to where 'I'is the typical rate of direct taxation -
adopt the aassumption that these effects can be neglected income plus any social security taxes - and u,, is
for most purposes with some form of special adjustment the relative income weight for urban workers
being introduced in those cases where the shadow wage normalised so that the relative income weight at
rate is clearly of critical importance to the acceptability of the critical consumption level is 1.0.
the project. 0 The new accounting ratios may be obtained by
substituting w f u l + A W * ~for~ w * , ~ in the
Adjustment procedure calculations specified in step 3 above.
Let me now summarise the results of implementing the 0 If the difference between the initial estimate of v * ,
adjustments discussed above within the context of the and the revised estimate of this parameter is
LMS method of estimating shadow prices.' I will do this significant, then the revised accounting ratios
by outlining the estimation procedure in terms of a series should be used to estimate the change in the
of steps. Unless otherwise specified accounting ratios primary input accounting ratios, Aw*, and the
should be understood as referring to constant price ARs, revision process repeated until the estimates
so that shadow prices will be the product of the accounting
converge. Usually it will not be necessary to
ratios and predevaluation prices: perform more than one further revision.
7. For accounting ratios to be used with post-
I. Following the LMS procedure outlined earlier,
devaluation prices, it is necessary to multiply v *, w *
calculate the input matrices A, B and thus the total
by I / ( I + ~ ) as in step 4 above.
primary input requirement matrix R = [I-AI-IB.
2. Calculate a set of current price accounting ratios w There are three particular problems which should be
for the primary inputs based on data relating to the borne in mind when implementing this procedure for
current, pre-devaluation, levels of wage rates, market adjusting shadow prices to take account of a prospective
prices and so on. devaluation. It is not possible to lay down any general
3. The accounting ratios for good and services may then precepts about them, so in the following discussion I will
be calculated from focus on the assumptions implicit in the usual approach
and on possible responses to the violation of these
V* = Rw*; W* = w (.I+T) (7) assumptions.
where q is the predicted proportional devaluation.
Labour markets. When estimating shadow wage rates we
4. A judgement must be made as to whether the
make a variety of assumptions about the relationship
devaluation will in fact affect relative prices and
between labour markets and the behaviour of wage rates in
incomes in the medium and long term. In part, this is
response to changes in demand. The usual model is based
an empirical question concerning the nature of
on the view that additional employment in the formal
complementary economic policies and the institutional
sector results directly or indirectly in the transfer of an
responses to the initial effects of the exchange rate
equivalent of workers out of small scale agriculture. This
change. In addition, it involves an assessment of
is certainly a convenient assumption, and it may even
whether real income or price changes are required in
represent the true response of the system reasonably well,
order for the devaluation to be successful.
though we have little evidence on which to base careful
5. If it seems probable that the devaluation will not or
tests of the hypothesis.
need not affect relative prices and incomes, then the
However, an exchange rate change may substantially
initial set of accounting ratios should be used to
affect the overall pressure of demand in various segments
evaluate all projects using constant prices. For
of the labour market. So long as a type of constant returns
projects assessed after the devaluation it may be
to scale assumption is thought to be appropriate this
necessary to use post-devaluation prices in which case
should not matter.
the relevant accounting ratios will be the initial ones
If, on the other hand, the non-agricultural sector is
multiplied by I ( I + T I .
almost as large as or even larger than the small scale
6. If it is believed that the devaluation will or must alter
agricultural sector in terms of employment, we might
relative prices and incomes, then the accounting ratios
expect such a change to be associated with an increase in
for post-devaluation costs and benefits will differ
the marginal product of labour in agriculture and a
from those relevant to the pre-devaluation period.
decrease in the ratio of workers moving out of agriculture
The change in the accounting ratios may be estimated
to new employment outside agriculture.
as follows:
In simple terms these two factors tend to offset each
0 Estimate the expected proportional fall in real other, but they imply that we ought to incorporate a fuller

118 Project Appraisal June 1986


Conversion factors and shadow exchange rates
analysis of informal sector urban labour markets when adjustment to the urban critical income level alone. This
evaluating the effect of the exchange rate change on the will imply an inconsistency between the urban and rural
social cost of labour. The same consideration applies if the critical income levels, which may worry some people.
devaluation is associated with a significant change in the However, I am inclined to believe that governments are
relative prices facing different sets of producers. For almost never consistent in their distributional judgements
example, the combination of lower real urban wages and as between urban and rural households, since typically
higher relative prices for agricultural export crops may urban dwellers receive much greater weight than might
significantly lessen the rate of transfer from agriculture to seem justified by their relative income level. If the critical
the urban sector in response to new employment income level is reduced by a proportion o , then the new
opportunities in the latter. income weights u^ will be given by:
It is, of course, easy to construct examples in which an
exchange rate change, or other policy changes, may affect u* = (I -$)EU (9)
the shadow wage rate. In practice, these are most likely to where E is the inequality aversion coefficient from an
be relevant to countries like Taiwan, Venezuela, Turkey isoelastic social welfare function (Scott I 976, Chapter 3,
and Mexico where the process of labour transfer out of or S-T, 136).
agriculture has already progressed a considerable way.
Even for such countries it is likely that the exchange rate Changes in input coefficients
change would have to be associated with quite radical
It has been shown that so long as prices continue to be
changes in policy, prices or labour market forces to alter
set in the same way, essentially on an input cost basis, we
the shadow wage rate (SWR) significantly.
can use the pre-devaluation input coefficiants, a,, brh to
This is because experience suggests that the
calculate accounting ratios and shadow prices for the post
predominant component of the SWR is the net social cost
devaluation situation. However, as we saw earlier, changes
of the additional consumption caused by extra
in the exchange rate may lead to changes in the social cost
employment. Certainly, in considering the implications of
composition of items subject to monopolistic or
a prospective devaluation serious thought should be given
discriminatory pricing in domestic markets.
to its effects on the assumptions underlying the SWR
Such changes are likely to be especially troublesome
estimates.
where the devaluation is associated with the removal of
However, it may prove more advisable to re-estimate
quotas or other import barriers. This is because the policy
the SWR at relative frequent intervals - say, every 3-4
changes may have the effect of eliminating monopolistic
years - rather than trying to make elaborate but probably
profits accruing to those controlling the import licences.
rather dubious adjustments to estimates based on current
Alternatively, they may switch the appropriate basis for
data.
assessing the social cost of an item from that of its social
Income weights. I have already stressed that a devaluation value in different uses to the social cost of increasing its
may require a fall in real incomes - probably urban real domestic supply. It is not really possible to suggest any
wages - if it is to be successful. This implies a fall in the general procedure for dealing with the latter case, except
social weight assigned to consumption relative to other to note that it will imply some revision of the input
uses of resources, at least as regards its embodiment in coefficients.
policy. In other words we should use lower income Arguably, in preparing shadow prices estimates using
weights when evaluating the distributional consequences of the LMS procedure one should always focus on the social
projects and the income benefits incorporated in estimates cost of increasing domestic supply, even when there are
of shadow wage rates, and so on. fairly rigid quotas, because marginal social productivity is
Strictly, it could be argued that the change does not extremely hard to measure sensibly and may be quite
affect the income weights themselves, but rather the social variable over time. In general, this seems a reasonable
cost and effectiveness of policies designed to reallocate basis on which to work but, in the case of particular
resources from consumption to investment and other uses. projects, it may be quite misleading to cost some key input
I do not find this a very plausible argument, because in on this basis, so that one must always be aware of the
almost all countries governments could implement taxes problem.
and other domestic policies to achieve this redistribution, Of course, these comments apply primarily to shadow
particularly with respectto urban real wages, without pricing items on the cost side of a project, since often on
resorting to exchange rate changes. the benefit side it will be appropriate to use some valuation
Thus, I would interpret a devaluation as implying a of the social productivity of the -project’s output(s) - such
reduction in the relative importance of consumption as water, electricity, roads.
combined with an intention to stimulate the overall level of Where quotas or other import restrictions lead to
utilisation of resources in the economy and to achieve a domestic prices which exceed the equivalent import parity
reallocation of their employment in production. Probably level, thus creating super-normal profits for importers or
the most appropriate way of incorporating the inferred others, the simplest approach is to introduce a special
decline in income weights is to assume that the critical primary input category: quota profits. Then, for as long as
income level has been reduced by the proportion 4 . the quotas are in force, a conversion factor appropriate to
Unless there are strong reasons for believing that real the social cost of these profits should be used for this
incomes in the rural sector will fall along with those in the primary input.
urban sector, it will probably be sufficient to make this After the devaluation and the elimination of the import

Project Appraisal June 1986


Conrwsiorr Jacrors atid shadow exchange r a m
restrictions this conversion factor will become zero - Hughes, G A (19791, Shadow Prices for Project Appraisal in
A W * ~=~ - w l q P- so that in this way the effect of the Morocco (mimeo: World Bank).
quotas can be integrated with the normal LMS estimation Hughes, G A (1980). Shadow Prices and Economic Policy in
Tunisia (mimeo: Cambridge).
procedure for accounting ratios and the method of Little, I M D 8 bhirrlees, J A (1974). Project Appraisal and
adjusting for the devaluation outlined above. In cases Planning for Developing Countries (Heinemann: London).
where this method would be inappropriate, it will be Little, I M D & Scott, M Fg (19761, Using Shadow Prices
necessary to adjust the input coefficients specifically in (Heinemann: London).
order to derive accounting ratios or shadow prices. Pursell, G (19781, Estimating Shadow Exchange Rates: The Ivory
Coast (mimeo: World Bank, Washington DC).
Scott, M Fg (1976). "A set of accounting prices for Kenya" in
Scott, MacArthur 8 Newbery, pages 3-232.
Scott, M Fg (1974). "How to use and estimate shadow
References
exchange rates", Oxford Economic Papers Vol26, pages 169-
Bacha, E & Taylor, L (1971). "Foreign exchange shadow prices: 184.
a critical review of current theories", Quarterly Journal of Scott, M Fg, MacArthur, J & Newbery, D (1976). Project
Economics. Vol 85, pages 197-224. Appraisal in Practice (Heinemann: London).
Balassa, B (1974), "Estimating the shadow price of foreign Squire, L 8 van der Tak, H G (1975). Economic Analysis of
exchange in project appraisal", Oxford Economic Papers, Vol Projects (John Hopkins University Press, Baltimore).
26, pages 147-168.
Bhagwati, J (19691, "On the equivalence of tariffs and quotas"
Notes
in J Bhagwati Trade, Tariffs and Growth, Massachusetts
Institute of Technology Press, Cambridge, Mass). 1. For a general discussion of investment with economies of
Dasgupta, P, Marglin S Et Sen, A K (19721, Guidelines for Project scale and a demonstration of the validity of the procedure
Evaluation (UNIDO, Vienna). outlined, see Hughes (1976).
Dasgupta, P & Stiglitz, J E (1974). "Benefit cost analysis and 2. This is the foreign price converted into rupees a t the current
trade policies", Journal of Political Economy, Vol 82, pages 1- exchange rate.
34. 3. Readers interested' in a more rigorous mathematical
Hughes, G A (19761, "Investment and trade for a developing treatment of the problem should refer to the author for an
economy with economies of scale in industry", Review of earlier version of this paper which gives the details of the
Economic Studies. Vol 43. pages 237-240. analysis.

I20 Project Appraisal June r986

You might also like