IMPORT DEMAND AND NON TARIFF BARRIERS
AN APPLICATION TO MOkOCCO
Giuseppe Bertola
(M.I.T.)
and
Riccardo Faini
Division Working Paper No. 1987-2
January 1987
Country Analysis and Projections Division
Economic Analysis and Projections Department
The World Bank
We would like to thank E. Grilli, N. Rossi and the participants to the
EPDCO workshop at the World Bank for very helpful comments and
suggestions.
The World Bank does not accept responsibility for the views expressed
herein which are those of the author(s) and should not be attributed to
the World Bank or to its affiliated organizations. The findings,
interpretations, and conclusions are the results of research supported
by the Bank; they do not necessarily represent official policy of the
Bank. The designations employed, the presentation of material, and any
maps used in this document are solely for the convenience of the reader
and do not imply the expression of any opinion whatsoever on the part of
the World Bank or its affiliates concerning the legal status of any
country, territory, city, area, or of its authorities, or concerning the
delimitation of its boundaries, or national affiliation.
ABSTRACT
The pervasive presence of quantitative restrictions on imports
in LDCs make the prediction of import response following trade
liberalization a particularly arduous task. It is true that
historically estimated elasticities could be of some guidance, even if
they did not allow for the impact of QRs, provided that we knew the
direction of their bias. In this respect conventional wisdom would
suggest that price responsiveness of imports should increase following
the liberalization measures. We show in this paper that this is not
necessarily the case and, under plausible conditions, the opposite will
be true. Drawing on the work of Neary and Roberts (1980), we analyse
the demand for imports where quantity restrictions cover only a subset
of a given commodity category. The model is then estimated on data
from Morocco. The results suggest that quantity restrictions had a
significant impact on both the level of imports and their sensitivity to
income and price variations. We also discuss some of the implications
of our findings for the trade liberalization process in Morocco.
TABLE OF CONTENTS
Page No.
INTRODUCTION ........... e................. .. ................. .1
THE IMPACT OF RATIONING.. ................................... 2
EMPIRICAL ANALYSIS: THE CASE OF MOROCCO .................... 6
(a) The Evolution of Trade Policy and Flows................6
(b) The Data...... ............** *e*.............. 10
THE RESULTS................................... . .... ...... 11
(a) Imports of Consumer Goods .......................... 11
(b) Imports of Investment Goods.................... .......16
CONCLUSIONS AND POLICY IMPLICATIONS.. ................... 20
REFERENCES...... .... e e ***......... ..............*........ 24
LIST OF TABLES
Table 1: Value Share of Imports Subject to Licensing .............. 9
Table 2: Tariff Barriers .......................... 10
Table 3: Imports of Consumption Goods (Linear
Expenditure System). ......... .... . .................. .15
Table 4: Imports of Investment Goods (Cobb-Douglas
Specification) ............... o .......... ..... 19
LIST OF FIGURES
Figure 1: Consumption and Investment Imports ............. 8
Figuer 2: Impact of Lifting Quantitative Restrictions............ o21
INTRODUCTION
It is by now widely agreed (Krueger, 1978: Khan and Zahler
1985) that imports flows will respond more rapidly to trade
liberalization than exports. This differential response is one of the
main justifications used by international organizations like the World
Bank to supplement structural adjustment measures with external
loans. Accurate prediction of import response to liberalization would
then appear to be essential to the design of a structural adjustment
package. Insufficient capital inflows would exacerbate any balance of
payments problems and may even lead to a later reversal of the trade
liberalization policy. On the other hand, an excessive inflow may
bring an unwelcome appreciation of the real exchange rate.
Unfortunately, the pervasive presence of quantitative restrictions on
imports in LDCs make the prediction of import response following trade
liberalization a particularly arduous task. Past evidence cannot be
relied upon to predict future import flows unless quantitative
restrictions are explicitly accounted for. However, non-tariff barriers
are usually treated in a fairly cursory function and even more accurate
treatments, like the one in Moran (1986), do not permit to isolate the
impact of QRs and recover structural behavioural parameters.
It is true that historically estimated elasticities could still
be of some guidance, even if they did not allow for the impact of QRs,
provided that we knew the direction of their bias. In this respect
conventional wisdom would suggest that price responsiveness of imports
should increase following the liberalization measures. We show in this
paper that this is not necessarily the case and, under plausible
conditions, the opposite will be true. Estimated elasticities, which
-2-
do not allow for the impact of QRs, are therefore an even worse guide to
future developments than previously thought. Drawing an the work of
Neary and Roberts (1980), we analyse the demand for imports where
quantity' restrictions cover only a Lubset of a given commodity
category. The model is then estimated on data from Morocco. The
results suggest -that quantity restrictions had a significant impact on
both the level of imports and their sensitivity to income and price
variations.
The paper is organized as follows: Section 2 presents a simple
m.odel of rationing. Section 3 provides a cursory overview of the
evolution of trade flows and policy in Morocco. Econometric
specification and results for consumption and investment imports are
presented in section 4. The last section offers some conclusions and
discusses a few implications of our findings for the trade
liberalization process in Morocco.
THE IMPACT OF RATIONING
We assume that there are two categories of imports, i.e. those
subject to licensing and those which can freely enter the country. It
is also assumed that the relative border prices among these two
commodities stay unchanged. While it would be desirable to dispense
with this assumption, available data would not allow us to derive two
separate price indices for rationed and non rationed commodities.
Consider the following cost function
c(V, PD,PM'PC)
-3-
where PDI PM and PC denote respectively the price of a domestically
produced commodity, of the freely importable foreign commodity and of
the foreign good subject to import licensing respectively while V is the
level of production (utility). We also have that
M= 0 PM PC C
where P� is the price of good i in the base year. Total real import
1
expenditure is equal to q = PC qC + PM qM . We want to assess the
impact of quotas on both total (q) and free (qM) imports as well as on.
the sensitivity of q and qM w.r.t. V and PM. Suppose that qC is subject
to a ration. The cost function under rationing is c* = c (V, PDI PM'
qc). From this we can derive a set of effective demand functions:
qi = q (V, P D PM, qC) i = D,M
Following Neary and Roberts (1980) we define as the virtual
prices the set PD' PM and PC which would induce the consumer to demand
exactly the ration level. It follows that
() q* (V, PD ) = q (v, PD M
(1 M D' PM' iC) M (V PD' PM' PC)
i.e. the rationed demand is equal to the unrationed one if the latter
is evaluated at the virtual prices. The following equation defines PC
(2) qC= qC (V, PD' PM, Pc)
Let us differentiate eq. (1) w.r.t. PM. Using eq. 2 and Young's
theorem, we have
Sq* a (45q /S5p 2
(3) M = M - -M C
SPM S M qC/6PC
where all derivatives are evaluated at virtual prices. To compare
notional and effective sensitivity of qM w.r.t. its own price we must
recall that in the unconstrained case P1M and P move together. We
therefore look at:
-qM (2qM .IqM) q q0/Sp1 + M qC/pC
SP1 SPM SPc SPc SqC/ SPc
Interestingly enough, this expression will be negative if the two
imported commodities are net substitutes (Sq /6p > 0) and own price
dominate cross price effects (Sqc/6p0 + Sqc /Spm<O). The implication
is that constrained elasticities are larger than their unconstrained
counterparts. This finding runs counter Le Chatellier kind of result by
Neary and Roberts., The intuition is however fairly clear. In the
unconstrained situation, when P1M increases, also PC goes up dampening,
if goods M and C are net substitutes, the negative effect on qM. There
is no such effect, of course, when the constraint is binding. As a
result notional elasticities are lower than constrained ones. It is
noteworthy that the Neary-Roberts result reappears if qM and qC are net
complements. As far as total imports are concerned, it is easy to show
that their sensitivity to P1 (i.e. to their common price 0) is again
greater in the constrained case under the same conditions as before.
Finally to find the impact of rationing on the sensitivity of qM w.r.t.
V (the output/utility level) we differentiate again eq. 1:
(5) . - - m -
6V 6V &P,C dq0/&PC
Suppose that goods M and C are both normal. An increase of V leads to a
higher demand for good M and, in the rationed case, amounts to a
tightening of the constraint. If the two imported commodities are net.
substitutes, the virtual price PC will increase causing a further
increase of the demand for good M. Therefore the effect of an increase
of V on qM will be more pronounced in the rationed case. Unsurprisingly
the opposite result holds for total imports.
For completeness' sake we only have to determine the impact of
price and income variations for Marshallian demands. As Neary and
Roberts show constrained and notional Marshallian demands are equal if
the latter are evaluated at virtual prices and at an expenditure level
appropriately corrected for the income effect of the ration:
(6) dM (P D' M 'Y PC . qC, qC) M[P D' M' C y+ ( C C) qC]
where y denotes nominal income. The effects of an increase in y are
easily traced. In the constrained case a higher y tightens the
constraints and, if goods M and C are rez substitutes, causes the demand
for M to rise further than in the unconstrained case. A similar result
holds for good D. We can infer that, given the adding up condition,
total imports must instead be less responsive to variations of income in
the constrained case. The results are less clear cut for uncompensated
-6-
price elasticities: gross substitutability/complementarity is relevant
in this case, and income effects blur the nicety of the previous
picture. An increase in the common price of free and constrained
imports influences the demand for unconstrained imports through two
channels: substitution away from good M towards the domestic good,
although, as discussed above, the degree of substitutability is less
than in the unconstrained regime; and, since purchasing the ration is
more expensive, income effect tending to decrease the demand for goods D
and M. The two move in opposite directions in altering the notional
price derivative of the freely impoi1ted good. The implication is that
no general result can be obtained when comparing uncompensated price
responsiveness 1- between the notional and the constrained cases.
EMPIRICAL ANALYSIS: THE CASE OF MOROCCO
a) The Evolution of Trade Policy and Flows
Since 1956, when the country gained independence, trade policy
in Morocco has been geared to the objective of fost ring
industrialization. However only in the first half of the seventies,
following the four-fold increase of phosphate price (the main foreign
exchange earner for Morocco), did a fully comprehensive effort to
promote the development of the industrial sector begin. The phosphate
price boom proved however to be short-lived, but even afterward the
1/ It is worth noticing that the previous results for price and income
responsiveness for constrained and notional demands cannot be simply
translated into statements about elasticities, insofar as import
quantities, which appear in the denominator for the expression of
elasticities, are by definition lower in the constrained case. Only
if effective derivatives are higher than notional ones can results
for derivatives be safely carried over to elasticities.
-7-
Moroccan authorities did not renounce their effort to promote growth
through higher public sector spending. The short run impact on growth
was positive, but associated to high and unsustainable external
deficits.. The latter reached in 197.7 16.5 percent-of GDP forcing the
government to take a set of measures aimed at reducing investai'ent
spending and restricting imports. Similarly, in 1983 when, due also to
a set of adverse external shocks, the current account worsened
considerably and a payment crisis erupted, Moroccan authorities had to
take both emergency and structural adjustment measures. A trade
liberalization program was implemented, based mainly on the reduction of
import duties, the phasing out: of import licensing, the pursuing of a
more activie exchange rate policy and the setting up of a more favourable
administrative environment for exporters.
Figure 1 depicts the behavior of consumption and investment
imports. The varying degree of the foreign exchange constraint can
explain to a large extent the cyclical variations of aggregate
imports. Superimposed to these effects are the long-run impact of
industrialization policies. The latter were paramount in reducing
imports of consumption goods to only 10 percent of total imports in 1985
from 17.8 percent in 1967, but also made the economy highly dependent on
imported capital goods (their share in total imports went up from 22.6
percent in 1973 to 29.2 percent in 1980). Trade policy was used to
translate both the short-run constraints and the long-term objectives
into actual import flows. Unsurprisingly both tariff and non-tariff
barriers were part of the tool-kit of trade policy. As far as non-
tariff barriers are concerned, since 1967 all imports are classified in
three groups. Commodities included in the first list (list A) can be
Figure 1: CONSUIPTION AND INVESTMENT IMPORTS
A) CONSUMPTION GOODS IMPORTS
1968.0 1970.0 1972.0 1974.0 1976.0 1978.0 1980.0 1982.0 1984.0 1986.0 1988.0
800. I E
I # E
OF## 0
|D #
700. | N H
l 8###*#
600. i N NNGNNN G N
NN#O N
| ADNNNB N
500. I N NN N
I BNNNNC
A
400. I
1968.0 1970.0 1972.0 1974.0 1976.0 1978.0 1980.0 1982.0 1984.0 1986.0 1988.0
B)INVESTMENT GOODS IMPORTS
1968.0 1970.0 1972.0 1974.0 1976.0 1978.0 1980.0 1982.0 1984.0 1986.0 1988.C
3.60 E E
N N
2.80I
N N H
2.001 5 N
NGNNN G N NK
1.20 I
# ### 0
I #A ND iN
#A #0 ##
I N# NCN
0.40 AD
1968.0 1970.0 1972.0 1974.0 1976.0 1978.0 1980.0 1982.0 1984.0 1986.0 1988.0
-9-
freely imported, commodities included in list B are subject to import
licensing and list C commodities cannot be imported except under special
conditions. Table 1 reports the ishare of total imports in Lists B and C
both at an aggregate level and for disaggregated functional
categories. This is apparently an adequate indicator of the importance
of non-tariff barriers. It steadily increases after 1972 following the
government policy aimed at speeding up industriAlization. It peaks
twice in 1978-79 and in 1983 when the deteriorating current account
balances forces Moroccan authorities to take emergency measures aimed at
curtailing imports and decreases in 1984-85 following the policy shift
toward trade liberalization.
Table 1: VALUE SHARE OF IMPORTS SUBJECT TO LICENS1N'J
Investment Goods Consumption Goods Total Imports
1970 22.4 22.3 30.4
1972 27.2 21.3 22.3
1978 39.7 57.6 59.0
1980 63.0 61.3 65.9
1982 23.7 60.0 57.7
1983 67.6 77.3 66.1
1984 24.7 60.0 17.8
1985 20.8 54.5 14.2
Tariff barriers follow a very similar pattern steadily
increasing until 1983 and declining afterward. Data on custom duties
are, unfortunately, very hard to come by and only snapshots at
particular points of time are available. It is useful, however, to
- 10 -
recall that, in the context of the liberalization program, the maximum
custom duty rate was set to 60 percent in 1984 and 45 percent in 1985.
Information is also available on two across the board tariffs, the
special.tax and the.stamp duty.. Both.-exhibit a clear upper trend until
1983 (Tabl'e. 2), with the special import tax declining afterward from 15
percent to 7.5 percent in 1985.
Table 2: TARIFF BARRIERS
Special Import Tax Stamp Duty
1970 2.5 1.0
1973 5.0 4.0
1978 12.0 4.o
1980 15.0 10.0
1982 15.0 10.0
1983 15.0 10.0
1984 10.0 10.0
1985 7.5 10.0
b) The Data
Econometric analysis has focussed on consumption and capital
goods imports. Both have been strongly influenced, albeit in opposite
way, by the trade policy orientation. Furthermore the recent surge of
consumption goods imports, following the implementation of a set of
til,ade liberalization measures, has to some extent reignited the
controversy on the benefit of less inward oriented policies.
A
Agricultural imports have been excluded from our analysis on the ground
that they are fully controlled by Moroccan authorities.
Value data for imports have been deflated by unit value indices
calculated by the Direction de la Statistique from a fairly
disaggregated basis. These indices have been taken as representative of
the CIF price of imports. To this we have added an indicator of the
tariff rate inclusive of the effect of the stamp duty, the special
import tax and indirect taxes. The omission of custom duties is not
serious for investment goods imports, as far as they benefitted from the
various investment codes exemptions but may somewhat false our results
for consumption goods. GDP deflators have been used as proxies of the
composite prices of consumption and investment goods. Finally domestic
consumption and total investments have been used as activity variables
in the two import equations. As far as the issue of determining an
indicator of quotas is concerned, we have relied on the value share of
imports subject to licensing. This is, recognisibly, a fairly imperfect
indicator. A decrease in it may even indicate a tightening of the trade
regime if it reflects the fact that licenses are less generously
granted. However this indicator appears to fit the evolution of the
restrictiveness of quantitative restrictions in Morocco quite well, as
we noticed earlier, indicating therefore that changes in the trade
regimes were mostly accomplished by shifting commodities across lists
and not by varying the restrictiveness in granting import licenses.
THE RESULTS
a) Imports of Consumer Goods
We consider the problem of a representative household which,
having solved the first-stage of its optimization problem, allocates
- 12 -
total consumption - among domestic and imported goods. Relative prices
are assumed to be constant within the two groups of goods. Some of the
imported goods are subject to (binding) quantitative restrictions.
We seek an empirical specification that makes it -possible to-
implement the theoretical results presented above using available data
for Morocco. Severe restrictions on the functional form are needed.
We selected the Linear Expenditure System as a compromise between
functional form flexibility and data requirements. 2/
Neary and Roberts (1983) show that the L.E.S. demand function
for an unconstrained good in the presence of binding constraints is
a.
(7) m. p. = yj p. + -J [E - pi Yi -I Pc Mc]
c
where y's are the subsistence levels of the goods among which total
expenditure E is allocated; a's are marginal propensities to spend the
super numeraire income; m's are levels of consumptions; indices i and j
refer to unconstrained goods, index c to constrained goods.
1/ Of course, total consumption is jointly endogenous; single-equation
estimation is carried out under the maintained assumption of
independence between random disturbances in the first-stage and
second-stage equations.
2/ We attempted to estimate a flexibLe functional form relating the log
of total imports to the log of expenditure and relative price of
imports, with coefficient assumed to vary linearly as a function of
the quota indicator. The results were unsatisfactory, mainly
because of high collinearity introduced by the presence of the
interaction terms (quota x price) and (quota x expenditure), with
quota, price and expenditure also entering the regression
separately. The same remarks apply to the estimation of the
investment import equation.
- 13 -
Under the assumption of constant (unitary by choice of units)
relative prices among domestically produced goods and among imported
goods, we can write the following expression for total real imports M:
(8) PM M = P (yf + Em ) + 1 - [E - PD YD - M (IYf + cm)]
where PM is the aggregate price of icmports, including tariffs; PD the
aggregate price of home goods; subscript f refers to freely importable
foreign goods, c to constrained imports, and D to domestically produced
consumption goods.
It is possible to check that the responsiveness of total
imports both to total expenditure E and to their own price decreases as
X becomes smaller and I becomes larger (more constraints are
imposed). Recall that we found that uncompensated price derivatives may
move either way between constrained and unconstrained regimes. In an
&m.
LES goods are gross complements ( 1 < 0) and this tilts the balance
spj
towards a positive sign for the uncompensated price response. On the
other hand, goods are net substitutes and own price dominate cross price
effect, so compensated price derivatir,es are larger in absolute value in
the constrained regime (in accordance with eq. 4).
We are interested in estimating required consumption and
marginal propensity to spend for the aggregate "total imports". Given
the assumed constancy of relative prices within the aggregate,
estimation would be straightforward if it weren't for the presence of
quotas. Lacking data on quota levels for individual goods, it is not
possible to estimate eq. (8) as it stands: some very stringent
assumptions are needed to make use in estimation of the available quota
information. In the light of the general features of quantitative
restriction policy in Morocco, we have tentatively assumed that:
- 14 -
6 /%' i (the ratio of marginal propensity to spend on
currently constrained goods to total marginal propensity to spend on
imports) is in every period closely approximated by the ratio of imports
subject to quotas to total imports, qt. This is not rigorously
defendable; it should, however, be close enough to reality to constitute
an improvement over neglecting quotas altogether.
lyf + Em is approximately constant (either all import quotas
are always set at or slightly above the required consumption level, or
at least when fewer restrictions are imposed the allowed amount is
increased, so that the above sum is close to constant).
With these assumptions, and allowing for partial adjustment
(strongly suggested by the data) we get from (8).
beta (1 -q )
(9) (Pm *) (1 - d) {alfa^P + t-- beta * *q [E -gamma PDt -
alfa * PMt]} + d (PM M)(t-
where alfa = I Yt + Imc beta =jai gamma = YD and d is the partial
adjustment coefficient.
Results of the estimation of eq. (9) by non-linear least
squares are reported in Table 3.
The estimates are rather precise and not unreasonable. The
dependent variables is imports of consumer goods, which are about 2.5
to 5 percent of total consumption expenditure E over the sample
period. The marginal propensity to spend is (notionally) higher than
that - about 9 percent. Estimated super-numerary income is positive
over all the sample, and the point estimates do not depend on the
- 15 -
Table 3: IMPORTS OF CONSUMPTION GOODS (LINEAR EXPENDITURE SYSTEM)
(standard errors in parentheses)
alfa 333.1 13 degrees of freedom
(94.27)
R2 = 0.997
beta 0.0897
C .030) DW = 1.58
gamma 9213.29 st. error residuals = 79.95
(1985.1)
out of sample forecasting error
d 0.643 (for 1985) is 2917.� - 3025
(0.115) Test stat. 1.81 (X1)
LM test2for auto-correlation
1.58 (X )
starting point of the iteration. Regressing squared residuals over
squared estimates of the gradient (which is a test for misspecification
if homoskedasticity is assumed, see White, 1981) gives X2(5) = 9.69.
While no formal Hausman test has been computed, non-linear two stage
(NL2S) estimates do not differ appreciably from the non-linear least
squares (NLS) ones (although this is probably due to the low number of
observations).
We also tried to estimate a standard L.E.S. demand function,
neglecting the presence of quantitative restrictions. The results are
extremely poor (very imprecise, and with a negative point estimate for
the marginal propensity to spend on imports), which is evidence that
quotas do matter and that our attempt to take them into account is not
meaningless, with all the arbitrariness of the assumptions discussed
above.
The results of the L.E.S. estimation can be used to roughly
figure out the impact on imports of an hypothetical lifting of all
- 16 -
quantitative restrictions in 1985. The L.E.S. predicts that nominal
imports would then have been 3837.83 million of dirham (MDH) in the
first year, and 6491.53 MDH in the long run. These very large effects
can be contrasted to-the relatively minor effect of a repeal of tariffs
with status-quo quotas. The tariff on consumer goods in 1985 has been
roughly estimated at about 10 percent for the purpose of this study; the
L.E.S. suggests that a reduction of the price of imports by that amount
(keeping quotas in place) would imply a nominal expenditure on imports
equal to 28.67.73 MDH in 1985 (about a 10 percent increase in real terms
over the status-quo forecast) and 3777.91 MDH in the long run (43
percent increase in real terms).
All these simulations keep total consumption expenditure and
prices of domestic goods unchanged, which of course is very much a
simplification. Still, they do seem fairly realistic and provide
insights as to the different effects of tariff and non-tariff barriers
to trade.
b) Imports of Investment Goods
Morocco, as many developing countries, is heavily dependent on
imported capital goods. Investment imports represented in 1984 31.7
percent of total investment, up from 25.8 percent in 1968. In modelling
imports of investment goods we consider a representative firm which
minimizes the cost of acquiring a given flow of new capital goods
subject to a transformation function between foreign and domestically
produced investment goods. A subset of foreign investment goods is
subject to quantitative restrictions. Finally relative prices among
foreign capital goods are assumed to be constant. Suppose that total
- 17 -
investment is equal to a Cobb-Douglas function of foreign and
domestically produced investment goods:
-v10 I = ID n If E c + +
DfeF fceC D ff
where I and ID denote respectively total and domestic investment, while
the index F (C) refer to those foreign investment goods which can
(cannot) be freely imported. The firm is assumed to minimize the cost
of acquiring I subject to eq. (10) and the quantitative restrictions.
The price of all foreign investment goods is assumed to be the same.
After substituting the first-order conditions into eq. (10) and solving
for If, we find that:
1 -s P 1-1,8 11E8c
(11) If( c (P) feF
cef D ieF,D ~ 8 e
where PM and PD denote respectively the price of foreign and domestic
capital goods. Then:
(12) In 1 If n lnZf+ 1 (lnI-U In I -(1 )ln (P /P)
feF f 8ac N M D
aD ln aD - f fln f)
where 8M 1 SD = a i (with i e F,C)
- 18 -
Let now assume that q, the value share of imports subject to licensing,
is equal to E a /E a. / Under the further assumption that all a's
are equal and rations are set at identical levels, after a little
algebra the estimating equation becomes:
(13) In I If = In (am(1-q)) + 1- q I ln I - (1-sM) ln (PI/PD) -
feF M1m qtMM
8M q ln (1I) - (1-$M) ln (1-6 - a m q ln ((8M q)}
where a is a constant which depends on sM and the number of imported
goods.
It is noteworthy that, in line with eq. (4), both output and
the compensated price elasticities increase with rationing, i.e., with a
higher value of q. It can be easily checked than in this problems free
and constrained imported goods are net substitutes and own-price
dominate cross-price effects.
The results of estimating eq. 13, with the addition of a lagged
dependent variable, by non-linear least squares are presented in Table
4. The measured price elasticity is .68 in the short run and quite well
determined. Adjustment to long-run levels is fairly rapid. The
statistical properties of the equation are satisfactory. The LM test
does not indicate at the 10 percent significance level the presence
of auto-correlation (X1 = 1.67). Similarly a general test of mis-
l/Again the assumption is not rigorously defendable. In the Cobb-
Douglas, it amounts to assume that the constraint is just binding.
- 19 -
Table 4: IMPORTS OF INVESTMENT GOOD (COBB-DOUGLAS SPECIFICATION)
(standard error in parentheses)
.32 14 Degrees of Freedom
(.10)
R2 = 99
a -1.25
(.52) st. error of residuals = .010
Lagged Dependent .10 Out of sample forecasting
Variable (.05) error (for 1985) is 7.123 -
7.1277
Test stat. .002 (X1)
LM test for auto-
c2rrelation:
Xi = 1.67
specification (White 1981) fails to detect, again at the 10
2 _1 2
percent level, any significant problem (X = 5.56). 1/ 2.
Finally the predictive power of the equation is more than satisfactory,
with projected nominal investment imports in 1985 equal to 6975 MDH
against an actual value of 7008 MDH. Unsurprisingly both the Chow and
1/ Had we run the test using only square terms and not the cross
products to save on ihe degrees of freedom, the results would not
have changed much (X3 = 4.57).
2/ However given the small sample size, the power of the previous tests
may be fairly low. There are at least two possible sources of
misspecifications. First total investment may be itself
endogenous. Second serially correlated errors will result, given
the presence of the lagged endogenous variable, in inconsistent
estimates. To allow for those two effects the equation was
reestimated by non-linear two stage least squares (NL2S). The
results are however fairly close to NLS estimates. The equation
again is well-behaved. The Anderson-Mizon test (a more appropriate
test for simultaneous equation systems) dyes not provide any
indication of structural instability (X1 = .48). Also a
regression of the residuals on their lagged values fails to detect
the presence of serially correlated errors (t14 = .74).
- 20 -
the Hendry tests take values which are not different from zero up to the
third decimal point.
We can now examine the impact of liberalization policies. If
the special import tax had been removed in 1985 the relative price of
foreign investment goods would have decreased by approximately 7.6
percent with real investment imports going up by 5.6) percent to 1513 MDH
from their actual value of 1433 MDH. The actual price elasticity (.74)
is higher than its notional value (.68) because of the presence of
quantity rationing. A similar exercise can be performed for non-tar�ff
barriers. Figure 2 compares the fitted values of imports from eq. 13
with the levels which we would obtain if all quantitattive restrictions
were eliminated. The importance of rationing effects is clearly a
function of our indicator q. The impact of removing quantitative
controls is fairly limited (at most 9 percent) until 1979 when more than
80 percent of investment imports become subject to controls. A lifting
of non-tariff restrictions in 1979 would have increased investment
imports by a hefty 20 percent. A.second peak in q occurs in 1983 when
due to the sudden payment crisis Moroccan authorities took emergency
measures to restrict imports. Almost 70 percent of imports were subject
to licensing, the abolition of which would have increased investment
imports by 17.5 percent.
CONCLUSIONS AND POLICY IMPLICATIONS
It may sound unsurprising to conclude by saying that trade
policies have a very significant impact on import decisions. Still
there is some use in it. We have seen how relative prices significantly
affect import demand. A 10 percent real devaluation would have reduced
... ~~~~~~~~~~~~~..... . , . .M .... . ..,..... . 3E+
Figure 2:- IMPACT OF LIFTING QUANTITATIVE RESTRICTIONS
ON INVESTIENT GOODS IMPORTS (PERCENTAGE
CHANGE)
0.200 I
F
N H
0.16 I I
0. 120 N
N N # N l
I # E
IC # #K
|F N # N |
0.0801 D;NN #N N N N
N N# N# H JN I
II N#E IN I
| NNANNNN8NNN #C #K
0 .040 I ##C#N
1968 1969 1970 1972 1973 1975 1976 1977 1979 1980 1982 1983 '985
in 1985 imports of consumption goods by the same amount in the short-run
and by 42 percent in the steady state. Even investment goods imports,
for which substitution possibilities appear from a priori ground more
limited would have decreased by 7.4 percent. The quantitative
importance of price effects means that tariff policy is not without
consequence on the actual import flows. However, the fact remains that
even a full abolition of the special import tax from say its 1985 level
would have only a relatively small effect on the real exchange rate
(with 30 percent custom duties, approximately 5.8 percent). The impact
on consumption and investment imports, given existing quantitative
-22-
restrictions, I/ would be equal respectively to 5.8 and 4.2 percent.
The numbers are significantly larger when the attention of the
liberalization process is devoted to non-tariff barriers. We have seen
how both consumption and investment goods imports would heftly react to
a lifting of quantitative barriers. The impact is particularly relevant
for consumption goods, for which our estimates indicate that, following
the repeal of quantitative controls, imports would almost double in the
long-run. The impact of QRs is not limited to import levels, but
extends also to price and activity elasticities. Our results suggest
that had QRs for consumption goods been lifted in 1985, their income
elasticity would have increased from .93 to 1.20. While the
significance of these numbers should not be overestimated, a more
reliable implication of our analysis is that estimated elasticities from
models which do not take fully into account QRs (including therefore
their effect on elasticities themselves) offer very little guidance to
predict the effect of trade liberalization. Furthermore we show that,
contrary to widespread beliefs, it cannot be stated blankly that the
price respcnsiveness of imports will increase when the impetus of the
liberalization process sweeps away the web of quantitative controls.
Our results can also be used to cast light on some relevant
policy trade-offs. Trade liberalization in Morocco has relied mostly on
tariff reductions. Our results show that the impact on the current
account is likely to be limited. However, the importance of tv-ie
taxes as a source of fiscal revenue points to a possible clash between
i/ Recall that changing the extent of QR's would also affect price
elasticities.
- 23 -
the need to stabilize the budget and the imperatives of the
liberalization process. As an alternative to tariff reduction, the
lifting of QRs may at first blush seem more promising. It would not
have a negative impact on the public budget. On the contrary, by
increasing import flows, it would have a beneficial effect on government
revenues. The other side of the coin is, of course, its .impact on the
current account, given the significant effect on import flows of
quantitative controls. The policy maker must therefore strike a very
delicate balance in selecting the pace and the tools of the
liberalization process. The simple message is that different
instruments have different and sometimes opposite and undesirable
impacts on the various policy objectives. This paper can move some
steps toward a better quantification of the involved magnitudes.
-2 4
REFERENCES
Burgess, D. (1974) "Production theory and the derived demand for
imports."- Journal of.International Economics,.-.105.-117..------ -
Goldstein, M. and Me Khan (1985) "Income and price effects in foreign
trade", ch. 20 in R. Jones and P. Kenen (eds.) Handbook of
International Economics, Elsevier Science Publishing.
Khari, M. (1974) "Import and export demand in developing countries", IMF
Sta,f;- papers, 678-693.
Khan, M. and Zailer R. (1985), "Trade and financial liberalization given
-external, shocks and inconsistent domestic policies", IMF staff
papers,, 22-55.
'., I'.
Krueg;er, A.' (1978) Foreign Trade Regimes and Economic Development:
Liberalization Attempts and Consequences, Ballinger Publishing
for NBER.
Moran, C. (1986) "'Import models for developing countries", EPDGL
" Working Paper No. 1986--S.
Neary, R. and K. Robertrs (1980) "The theory of household behaviour
under rationing, European Economic Review, 25-42.
White, H. (1981) "Cnsmequences and detections of misspecified non-
linear regression models", Journal of the American Statistical
Association, 419-433.
..I , 4.
/