This volume contains appendixes that provide data and methodology supporting the analysis in Volume 1 of the study "Trade, Exchange Rate, and Agricultural Pricing Policies in Brazil". Appendix A includes tables showing effective protection rates for different sectors in Brazil from 1967-1980, and subsidies for Brazilian manufactured exports. Appendix B contains data on domestic resource costs and export subsidies for Brazilian manufacturing groups. Appendix C presents methodology and data on measures of protection for Brazilian agriculture. The remaining appendixes contain additional data tables and figures supporting the analyses in Volume 1.
1. 10185
WORLD BANK VOL. 2
COMPARATIVE STUDIES
The Political Economy of Agricultural Pricing Policy
Trade, Exchange Rate, '
and Agricultural Pricing Policies
in Brazil
VolumeII Appendixes: Data and Methodology
Antonio Salazar P. Brandao
Jose L. Carvalho
~ _
p.~~~~~~~~~~~~~~~~o
FL'O
2.
3. /t1 II{
The Political Economy of Agricultural Pricing Policy
Trade, Exchange Rate,
and AgriculturalPricing Policies
in Brazil
VolumeH Appendixes: Data and Methodology
Antonio SalazarP. Brandao
Jose L. Carvalho,
WORLDBANK
COMPARATIVESTUDIES
The WorldBank
Washington,D.C.
5. iii
Abstract
This study covers the latter part of the 1960s, the entire 1970s,
and the first years of the 1980s.During this period, agriculturalpolicy
has undergone major changes, going from a period in which subsidized
credit to agriculturewas the most importantpolicy instrumentto one in
which guaranteed prices assumed a predominant role. In addition, as
Brazil's economic situation became more and more unstable, so did
agriculturalpolicy.
A large number of government interventions occured in the
agriculturalsector of Brazil, and both direct and indirect intervention
helped to distort prices. To evaluate the net effect of this array of
policies, various measures of protection were calculated.The results
indicate that, in general, the agriculturalsectqr was taxed in Brazil
when compared to the nonagriculturalsector. Two export crops, soybeans
and cotton, were taxed more heavily than import-competingfood crops.
Indirect causes, mainly exchange rate overvaluation, were the most
importantdeterminantsof the level of taxation.
Substantiallossesof productionwere estimatedas a consequenceof
the discriminatorypolicies of the government.The overall impact on the
trade balance was also significant.
Although the agricultural sector was taxed through price policy,
this becomes less clear when all transfersto and from agricultureare
accountedfor. The most importantof these transfersduring a large part
of the 1970s and the early 1980s was that associatedwith a subsidized
interestrate for agriculturalcredit. The data make it clear that what
the governmentdid over that periodwas to use creditsubsidiesto counter
balance its discrimination against agriculture. This compensation,
however,was highly regressivewith respectto incomedistribution.Price
discriminationaffectedall agriculturalproducers,but the creditsubsidy
was not available to all of them. It was the larger and wealthier
producers that received the transfersassociatedwith that policy. This
contributedto the concentrationof incomeobserved in the agricultural
sector during the 1970s. Our assessmentconfirms that smaller producers
paid the largestpenalty for price discriminationagainst the sector.
Inflationwas clearlyone importantconcernof the governmentin its
price interventions,especially with respect to food crops. Pressure
groups were particularly active for soybeans and cotton, which are
importantindustrialinputs. However,a high degree of randomnessin the
behavior of the governmentis also seen in the econometricresults.This
was particularly relevant for those commodities associated with many
pressure groups with possibly conflictinginterests.
6.
7. - v -
Table of Contents
APPENDIX A TABLES................................................. 1-16
APPENDIX B ........................................................ 17
APPENDIX B TABLES................................................. 25-30
APPENDIX C ........................................................ 31
APPENDIX C TABLES .. ......... 40-47
FIGURE C.1........................................................ 48
APPENDIX D........................................................ 49
APPENDIX D TABLES .................. 51-60
APPENDIX E........................................................ 61
APPENDIX E TABLES .................. 63-77
APPENDIX F........................................................ 78
APPENDIX P TABLES ................ 83-103
TABLES 4.10A TO 7.4A .................. 104-125
TABLES A.7.1A TO 8.6A .................. 126-137
TABLES H.1 TO H.27 ................... 138-164
8.
9. Table A.l - EFFECTIVE PROTECTION RATES IN BRAZIL, 1967 - 1980 (percent)
SECTOR 1958 1963 1967 1973 1975 1980/81
Primary vegetable products - 47 - 15 - 14 na na na
Primary animal products 24 12 na na na na
Mining - 5 34 9 - 8 - 8 - 4.2
Mtanufacturing (average) 106 184 48 27 30 46.4
Nonmetallic Minerals 73 103 48 21 26 - 19.6
Metallurgy 61 124 33 18 17 34.2
Machinery 22 68 31 9 13 93.3
Electric Equipment 83 169 57 19 22 129.3
Transport Equipment 82 147 81 30 37 . 6.5
Lumber and Wood 138 176 44 19 24 17.7
Furniture 221 367 92 44 42 52.7
Paper and paperboard 86 169 42 24 32 - 18.5
Rubber products 139 221 182 56 54 - 21.4
Leather products 248 405 84 26 36 13.9
Chemicals 56 146 20 28 22 86.4
Pharmaceutical products 17 60 10 28 17 116.3
Perfumery 279. 453 70 33 40 91.6
Plastics 281 489 117 99 111 28.3
Textiles 239 298 88 36 58 36.7
Apparel and shoes 264 481 154 26 37 46.7
Food 502 678 71 33 37 26.1
Beverages 171 243 76 143 139 - 1.1
Tobacco 273 469 79 - 6 - 6 5.7
Printing and Publishing 139 305 8 10 13 31.9
Miscellaneous 88 175 45 17 21 171.7
Average 30 75 14 25 29 na
Source: 1958-75 Carvalho-Haddad (1980); 1980/81 Tyler (1981)
t;.rv:(a) Effective rates from 1958-1967 were computed by Fishlow; those for 1973 and 1975 were computed by Neuhaus-Lobato (1978)
and those for 1980/81 by Tyler (1981).
20. - 12 -
TABLE A-I0 - BRAZIL: Monthly Minimum Wage in Rio de Janeiro - 1955-85
MINIMUM WAGE
MW APPROVALDATE CurrentCr$ May 1977Cr$
JUL 1954 2.40 1,628.53
AUG 1956 3.80 1,700.13
JAN 1959 6.00 1,813.99
OCT 1960 9.60 1,701.81
OCT 1961 13.44 1,722.51
JAN 1963 21.00 1,573.59
FEB 1964 42.00 1,504.37
MAR 1965 66.oo 1,242.10
MAR 1966 84.oo 1,138.70
MAR 1967 105.00 1,051.21
MAR 1968 129.60 1,076.83
MAY 1969 156.00 1,018.91
MAY 1970 187.20 998.44
MAY 1971 225.60 987.63
MAY 1972 268.80 1,004.01
MAY 1973 312.00 1,027.17
MAY 1974 376.80 975.05
MAY 1975 532.80 1,099.19
MAY 1976 768.00 1,109.03
MAY 1977 1,106.40 1,106.40
MAY 1978 1,560.00 1,147.04
MAY 1979 2,268.00 1,143-56
NOV 1979 2,932.80 1,105.91
MAY 1980 4,149.60 1,150.94
NOV 1980 5,788.80 1,167.03
MAY 1981 8,464.80 1,145.06
NOV 1981 11,928.00 1,161.87
MAY 1982 16,608.00 1,125.28
NOV 1982 23,568.00 1,168.99
MAY 1983 34,776.00 1,089-74
NOV 1983 57,120.00 1,029.50
Source:IBGE- AnuarioEstatisticodo Brasil.
aObtainedas [MW(t)/CPI-RJ(t)]CPI-RJ (May1977)whereMW (t) is
theminimumwageat periodt; CPI-RJ(t)is the ConsumerPrice
Indexin Rio de Janeiroat periodt.
25. - 17 -
APPENDIX B
PRICE INDEX CONSTRUCTION
B.1 - DOMESTIC PRICE INDEX OF IMPORTABLES (PM)
The Price Iftdexof Importables was obtained as an
weighted average of agricultural importables (PMA) and non-
-agricultural importables (PMNA). The weights for PMA, the a i's
were the participation of wheat imports on total imports. The
PMNA weights were defined as (1 - a.i. Although other agricultural
products are imported occasionally, we think wheat is a representativ
product of agricultural imports.
PMNA from 1970 onward is given by the wholesalepriceson
cereals and grains (column 19) as computed by Getulio Vargas
Foundation (FGV).Prior to 1970we used the wheat price paid by the
millers.
PMNA from 1960-69 corresponds to the aggregation of two
wholesale price indices computed by Vargas Foundation: chemical
and metalurgy aggregated by theircorresponding participations on the
sum of their values of production. For the rest of the sample period
that is, from 1970 on PMNA is obtained by aggregating the following
wholesale price indices (also computed by FGV): Lime and Silicates,
Nonferrous Metals, Machinery, Electrical Machinery Equipment,
Vehicles. and Vehicle Equipments, Chemical, Paper and Paperboard,
Artificial Yarn Fabrics. These indices were selected to define
non-agricultural importables based on Carvalho-Haddad (1981). Using
26. - 18 -
a four digit disaggregationof the industrialsector in Brazil,
these authors classified them on exportables,importablesand home
goods based on an statistic that depends on production, consumption,
exports and imports. For details on this classification,readers
are to refer to Chapter III and table A.8 of the referred study. The
aggregation of these indices on PMNA from 1970 on, was made weiaht-.nc
each index by the relative share of the value of production of the
corresponding sector on the sumof the considered sector values of
production.
B-2 - DOMESTIC PRICE INDEX OF EXPORTABLES (PX)
Once more, an agriculturalprice index and a non-agricul-
tural one were computed andaggrea-atedtodefinethePX. The agricul-
tural price index of exportables (PXA)from 1970 on is the index, at
thewholesalelevel,of AgriculturalProducts mainly for Export. Prior
1970 we used the wholesale price index of agriculturalproducts.
The price index of non-agriculturalexportableswas
obtained by aggregatingwholesale price indices computed by FGV as
follows: 1960-1969 - Raw-Materials;Hides, Skins and Leather; Textiles
Apparel and Footwear;
1970 onward - Lumber and Wood Products; Furniture; Iron and Steel;
Hides, Skins and Leather; Natural Yarm Fabrics;
Hosiery; Apparel (hosieryexcluded); Footwear.
These indices were aggregatedon PXNAby weighting each
one by their relative importanceon the group as measured by the
value of production activity participationon group total value.
27. - 19 -
To aggregate PXA and PXNA to obtain PX, we used as
weights the relative participation on exports of agricultural
products for the first and its complement to one for the second.
B.3 - DOMESTIC PRICE INDEX OF NON-AGRICULTURAL PRODUCTS (PNA)
To construct a domestic price index of non-agricultural
products, we aggregated the following price indices: non-agricultural
importables (PMNA) as defined in A above; non-agricultural exports
(PXNA) as described in B and non-agricultural domestic goods. The
price index of non-agricultural domestic goods (DNA) was cbtained
by aggregating the following indices-computed at consumer level in
Rio de Janeiro by FGV: Housing, Personal Services and Public Services
These indices were aggregated in DNA by their relative importance
on the RJ-CPI, normalized to add up to one.
To aggregate PMNA, PXNA and DNA to obtain PNA we did
the following:
PNA = a (M) PMNA + a (X) PXNA + a (D) DNA
where:
MNA IPa (M) = ( A ) .( _ )
MNA+ XNA PIBNA
)(X) = (XNA IP
MNA + XNA * PIBNA
a (D) = -PIBNA
28. - 20 -
MNA = non-agricultural imports in current cruzeiros;
XNA = non-agricultural exports in current cruzeiros;
IP = industrial product in current cruzeiros;
PIBNA = non-agricultural product in current cruzeiros;
SP = services product in current cruzeiros.
B.4 - DECOMPOSITION OF CONSUMER PRICE INDEX ON TRADABLES AND
NON-TRADABLES
The decomposition of the CPI on tradables and non-tra-
dables was done considering as non-tradables the following components
housing, personal services and public services. These components are
computed monthly for Rio de Janeiro city by Getulio Vargas Foundation
To aggregate these components in the non-tradable price index we
--roceeded as describedpreviously for DNA (B.3 above). Thus, non-tra-
dables price index do not include agriculturalproductsand therefore
is the same index as domestic non-agricultural products. This might
not represent a serious bias in this index since very few agricultu-
ral products, such as cassava, can be considered domestic products.
The tradable component (PT) of the consumer price index
was obtained by residual considering the sum of the weights of the
individual components that comprises the non-tradable price index as
the weight of this price index (PNT) in the definition of CPI. Thus,
PT was obtained as:
PT 1 CPI a( ) PNT,
at a
29. - 21 -
where a had the following values:
Period _
1960-65 0.637
1966-71 0.706
1972-73 0.671
1974-84 0.621
according to the family expenditure surveys conducted by FGV to
calculate CPI for Rio de Janeiro.
The reason for using Rio de Janeiro CPI and not a
national one is that this index became available from 1979 on.
Nonetheless, the relative movements of both series for the
overlaping period is quite similar and therefore the Rio de Janeiro
CPI can be used as a good proxy for the national index.
30. - 22 -
B.5 - Undistorted Price Index for Non Agricultural Goods
To calculate the free-trade price of non-agricultu-
ral goods, UPNA, the following procedure has been adopted:
E* PXNA + (M) E* PMNA
UPNA = a (X) . E . + a (M) + a(D) DNA
X m
Where, sx and tm are export subsidies and import
tariffs on non-agricultural goods; PNXA, PMNA and DNA are the
components of PNA as defined in B.3 above as well as a (X), a (M)
and, a (D); E is the official exchange rate and E* is the
equilibrium rate as estimated in the Appendix C. This procedure
is incomplete in correcting PNA since distortions on tradables do
affect non-tradables (DNA). One way to correct DNA would be to
consider the equivalent homogeneous tariff as the price distortion,
weightedby the cross-price elasticity of domestic goods with
respect to tradables that in this case shouldbeapproximated by w
defined in Appendix C. For simplicity and given the difficulties
associated to obtaining this cross-price elasticity we did not
consider here this correction.
The estimates of the export subsidies come from
various sources: Musalem (1981) for the period 1964-78; De la Cal
(1981)for 1979;Tyler (1983)for 1981. For the remaining years, that
is 1980, 1982 and 1983 we estimated these export subsidies based
on observed price,indices as follows:
P
(1 +s)s x where
x P*
x
31. - 23 -
P* - Export price index as calculatedby GetClio Vargas Foundation;x
Px = Domestic price index of exported goods as calculated on
table B.2 above. To maintain the consistenceof the series,
we computed the rates of change of sx estimated as above and
apply them to the previously mentioned estimates.
For the import tariffs, the adopted estimates also
come from various sources: Fishlow (1975)for 1962; Fendt (undated)
for 1977; Guedes et al. (1981) for 1973, 1979 and 1980; and Tyler
(1983)for 1981. To obtain the missing year we adopted a procedure
similar to the one adopted for sx defining import tariff as follow:
p
(1 + t) m
m
m~~~~~~~~
-m Import price index as calculated by GetuilioVargas Foundation;
PM - Domestic price index for imported goods as calculated on
table B.1 above.
One could argue that using our definition of x
and tm one would have a good estimate of sx and tx. We did
calculate these subsidies and taxes. The results we obtained
were quite inconsistentwith the existing evidences. This is
the case for two reasons. First the definitionswe presented
above are valid for individual prices and might not be a good
estimate for aggregates. Second the price indices we constructed
do not cover the same aggregates used by Getulio Vargas Foundation
in constructingP*x and P*m indices.
32. - 24 -
B.6 - Undistorted Consumer Price Index
The undistorted consumer price index can easily
be obtained by the decomposition we made on this index on B.4
above:
CPI = aPT + (1 - a) PNT,
where
PT is the price index for tradables and PNT is the price index
for non tradables.
Thus, tradables price index could be corrected by
taking into consideration the average tariff described in the
Appendix C below and the exchange rate correction. Also, PNT
could be corrected in the sane fashion as DNA above. For simplicity
and because the most relevant distortion here is reflected on
the exchange rate, we adopted the following correction for CPI.
UCPI = E aPT + (1 - a) PNT
E
The values for UCPI are reported on Table B.6 for
two alternative equilibrium exchange rate.
38. - 30 -
TABLE B.6 - BRAZIL: Consumer Price Index Correctedfor Exchange Rate
Distortionon Tradable Goods
C P I - CORRECTED
YEAR C P I - RJ
E*wb 80
1960 0.51 0.571 0.529
1961 0.68 0.732 0.695
1962 1.03 1.162 1.073
1963 1.76 2.243 1.744
1964 3.37 3.812 3.327
1965 5.59 6.686 5.372
1966 7.90 10.932 7.972
1967 10.31 15.040 10.723
1968 12.61 17.427 13.393
1969 15.39 21.038 15.978
1970 18.88 26.376 19.894
1971 22.69 32.086 24.923
1972 26.45 36.932 28.581
1973 29.80 41.024 31.691
1974 38.03 50.139 44.170
1975 49.04 63.967 55.824
1976 69.57 90.881 76.676
1977 100.00 132.913 106.283
1978 138.70 186.339 151.110
1979 211.80 268.070 235.987
1980 387.20 427.670 427.283
1981 795.90 906.030 867.582
1982 1 595.70 1 771.564 1 794.304
1983 3 812.90 3 519.219 4.146.411
Notes: CPI - As published by Getulio Vargas Foundation for Rio de
Janeiro citv
CPI - Corrected - See B.6 above.
E*wb - Equilibrium Exchange Rate obtained by PPP criterion
base year 1980 as computed by the World Bank as in
table C.S.
80 - Equilibrium Exchange Rate based on the year by year
elasticity approach as described in appendix C table
C.4. Base year 1980.
39. - 31 -
APPENDIXC
EQUILIBRIUMEXCHANGE RATE
Following the note prepaired by Schiff (January23 1986) the
equilibrium exchange rate is defined as:
E*= (ltmQdnQs)t
[Q + Q + n Q x °
+~~~~~+IE
Od = Demand for foreign exchange;
as = Supply of foreign exchange;
Eo = Official exchange rate
600= Qd - as at the official exchange rate Eo
E* = Equilibrium exchange rate at the moment E0 is observed;
n = Price elasticityof foreign exchangedemand (positivelydefined)
E = Price elasticityof foreign exchange supply;
tm = Tariff on imports;
tx = Export taxes
This formula.was applied to Brazilian data with
tm and t, estimated as described below. However, the results were
not satisfactory,specially in the years of the beginning of the
period: for example, in 1965, the exchange rate estimated according
to this formulawas 1.584 and the official rate was 1.891, which
indicates that the cruzeiro was undervaluedby approximately19%.
40. - 32 -
In view of these difficulties, the above
formula was not utilized. Instead, we have estimated the
equilibrium rate following the procedure of Roe and Green (1986),
which consists of the derivation of an exact formula for E*
assuming demardand supplyof foreign exchange subject to constant elas
ticities. The estimates are shown in Table C.4 of this Appendix.
They are, in general very similar to the ones obtained before
with the formula given in the beginning, but for the specific
years in which those results were not very satisfactory it
improves them. The year 1965 shows an undervaluation of only
5.89%.
The procedure of Roe and Green is
described below. Let the excess demand for goods and services
in period t, Mt, be
Mt = Bt(p = Bt Et P* (1 +
where Bt is an intercept term that reflects the movements in other
variables affecting Mt; Pmt is the domestic price of imported
goods and services and is equal to world (or border) price p*,
times the exchange rate, Et, times one plus the tariff (l+TMt).
The value of imports is then:
pt Mt BdtBt (P* )l+7 Et (1+Tmt)]
Similarly, the supply of foreign exchange
can be obtained. Let Xt be the excess supply of goods and
services, P*xt be the world (or border) price of export goods
and services and Txt be the export tax. Then, it is assumed that
41. - 33 -
Xt = At pxPtEt (1-txt)]O
from which one obtains the value of exports;
P* Xt = = At (P )1+a EEt (l-txt)]a
where At is, again, an intercept term that reflects changes
in exogenous variables that affect Xt.
Data on the value of imports and exports
of goods and services, on Tmt and Txt,Et and the elasticities
a and n allow us, following Roe and Green, to estimate At and
B . The equilibrium exchange rate is defined, for each t, as
the value of Et thatonewouldobtainunderthe followingconditions:
l Mt :xt'
T
xt = Tmt = 0 and At and Bt as estimated above.
To compute the equilibrium exchange rate
we will need the true tariff and the true subsidy accruing on
imports and exports. We will also need the demand and supply
elasticities associated to foreign exchange and a set of published
data on trade. Let us consider the required information by turns.
42. -34-
Calculating tm and sx
Accordinglywith Sjaastad-Clements(1981) true
import tariff (t ) and true export subsidy (s ) can be defined as
function of two parameters, w and t, as follows:
=m 1-w)t
lwE
X +wt
sx I + wf
where
w - distributionalparameter: elasticity of (homegood prices/export
prices) with respect to (import prices/export prices);
-- the uniform equivalent tariff.
Note that if sx < 0, that is t > 0, there will be a true tax on
exports. This is so because w has to ve positive as shown by
Sjaastad-Clements (1981).
Hence, if we have estimatesof t and w, we will
be able to estimate tm nd s
An estimate of t
For exportablesand importableswe should have:
Px E . (1 + s9) Px (c.1)
Px E * ( + tm) P
43. 35 -
where
Px = domestic price of exportable goods
P* = international price of exportable goods (in US$);
x
E = official exchange rate (Cr$/US$);
Pm = domestic price of importable goods;
Pm = international price of importable goods (in US$).
m
(c.l) should hold true for every tradable. Since these individual
price comparisons cannot be made for a long period of time due to
data limitations, the comparisons we will make will consider price
indices for each tradable category. Thus,
v 1 + t P*
= +tm m- (c.2)
P 1 + s x
x x x
will hold for the corresponding price indices.
If we call 1 + tm (1 + t), E is in fact the equivalent
1 + s,
uniform tariff.
Since we are working with relative price indices
and not with the individual prices we cannot estimate (1 + Et)by
taking (P /P )/(P*/P*). Nonetheless, we can obtain the rate of
m x m x
change in (1 + t) by applying d lg to the expression (c.2).
P p*
d lg m d lg (1 + t) + d ig m
x P or
x ~~~~~~~~~x
d lg (1 + t) = d lg pm _ d lg (c.3)
x x
44. - 36 -
Estimating a value of t for one year, will
allow us to obtain the series of t once d lg (1 + t) is
estimated by (c.3). Pm and Px can be taken as the price indices
constructed on appendix B while Pm and Px are published by Getuilio
Vargas Foundation. Therefore, d lg (1 + t) can easily be obtained
from (c.3) and the next step is to compute t for a specific year.
Tyler (1983), using 676 different products,
distributed along 72 sectors of the 1970 IBGE input-output table,
calculated the average implicit nominal tariff protection and the
effective rate of export promotion. These values will be taken as
proxies for tm and sx respectively. Tyler's estimates are based
on data collected along the period June 1980 - April 1981. Thus,
we will take his results for the year of 1980.
Table C.1 - BRAZIL: 1980 Estimates for tm and sx
Sectors t sm x
Agriculture - 7.2 - 5.4
Transformation Industry 24.5 34.9
Capital Goods 45.5 34.9
Intermediaty Goods 25.2 34.7
Consumer Goods 13.1 35.0
Weighted Average* 21.62 18.06
Source:Tyler (1983).
*Weightsdefinedas importparticipationfor tm and exportparticipationfor sx
With the average values for tm and sx for 1980, we
can compute (1 + E) as follows:
45. - 37 -
+ tm 1,2162
(1 + t) 1 + 8 = 1,1806 = 1,0302.
Thus, the value of t for 1980 is 0,0302. With
the values of d lg (1 + t) computed by (C.3)we can obtain a series
of t as reported on table C.2.
w Estimates
Fendt (1981),estimatedw, following
Sjaastad-Clements (1981) through the simple regression.
19(d) a + b 19 + c 19(Z ) + u,
tAx ) X(Px)
being w defined as
d lg (Pd /Px
where
d lg (P /P
pd = Price index for domestic goods
dx= Domestic Price Index for export goods
P = Domestic Price Index for import goods
As Fendt estimated, w depends on lg (Pm/Px).
Using the average values of this relative prices for the
correspondingperiod, he estimated the followingvalues for w:
1955 - 59 w = 0.69
1974 - 78 w = 0.43
Since we could not reproduceFendt's Pd, px
and Pm' we use our price index series from Appendix B to estimate
w for different periods using the following regression.
46. - 38 -
lg ( d) = a + w lg X +
The results we obtained were:
1964 - 73 w = 0.57 R = 0.64
2
1974 - 83 w = 0.43 R = 0.69
Given the results presented by Fendt and our
own estimates of w, we decided to adopt the following values
for this parameter:
Period w
1960 - 63 0.69
1964 - 73 0.57
1974 - 83 0.43
Thus, with these values for w plus the estimates
of t discussed above, we computed tm and sx. Since the export
subsidies were negative for all periods, they were in fact export
taxes. In table C.2, we report the estimates of w, t, tm and tx
from 1960 to 1983. Alternatively,we present on Table C.2a the
estimated values for tm and tx consideringthe 1976 value of t
estimated by Fendt (1981). In this way, the two sets of tm and tx
will be used in generating the equilibriumexchange rate.
The hypothesis on the elasticities
The methodologyused here to compute the
equilibriumexchange rate requires . knowledge, of the price
elasticitiesof demand (n) and supply (e) of foreign currency
(Us$ in the Brazilian case). Since we have no reliable estimates
47. - 39 -
for these elasticities we decided to adopt the following values:
n = 2.0 and c = 1.0
Several studies on the equilibrium or shadow
exchange rate have been done for Brazil. Those that have used
the elasticity approach had to impose some hypothetical values for
n and E-L/. In general, their sensitivity analysis induced those
authors to choose values for these elasticities close to the ones
we are adopting here.
Using the trade data reported on table C.3 plus
the estimated values of tm and tx and the hypothesis about n and e,
we computed the equilibrium exchange rate. This rate is also
computed by the values of t obtained from Fendt estimate for 1976
and is reported on table C.4.
Another way of computing a series of the
equilibrium exchange rate would be as follows: take an estimate for
the equilibrium exchange rate calculated for one year and apply
the principle of constant purchasing power parity to obtain the
series from this calculated value.
We have for Brazil at least four points in time
for which we have estimates of the shadow exchange rate. Bergsman-
-Malan (1970) estimated this rate to be Cr$ 0,22 per US$ dollar in
1960.
11 See for example Bergsman-Malan (1970),Oliveira (1981) and a World Bank
Office Memorandum of December 28 1977 prepaired by F.J. Easwaker and
P.T.Knight.
52. - 44 -
Easwaker and Knight on their World Bank
Memorandum of December 1977 reported an earlier study where they
estimated that in 1975 the shadow exchange rate was 25% above
the official rate. For 1977 they reported that this difference
was increased to 30%. Another World Bank mission, updated this
estimate to 1980 and found that the shadow exchange rate was 16,7%
above the official rate.
In this way, using the constant PPP argument,
we can arpute four series of equilibrium exchange rate. Nonetheless,
we will compute only two. One based on Bergsman-Malan and the
latest estimate done by the World Bank that is for 1980.
The problem with the PPP aproach is that any
change in relative prices are not considered. Therefore, the
equilibrium rate thus computed will lose its meaning as the
relative prices do vary. This is exactly the case for the considered
period. In this way, we intended to use the difference between-
the equilibrium rate based on the 1960 estimate (E*0) and the
equilibrium rate based on 1980 estimate (E*0) to infer about the
relative price changes. In this way, (E*0 - E*0) along the period,
could be explained by the real change in the relative price of
tradables. With this information, the equilibrium rate based on
PPP could be corrected account for relative price changes.
Table C.5 reports the official rate and those
equilibrium rates computed by the PPR approach taking as base year
1960 (Bergsman - Malan) and alternatively 1980 (World Bank). As
one can easely note by inspecting this table, the differences
between these two rates are negligible and therefore no attempt
53. - 45 -
to correct for relative price changes can be made.
This is a quite striking result since along
the considered period the relative prices did change at least
after 1973 with the first world petroleumprice shock. Once again,
we should expect another drastic movement in these relative prices
in 1979 with the second petroleum price shock. Also, from 1972
through about 1977 the commodity prices did present a real upward
trend. In any case, it is difficult for us to accept that the
terms of trade for Brazil remained stable during this period.
56. 20
10
0
-20
-30 I II /
1460 19'63 I 56 1-9169 -- - 19-t2-- 1075-'- -- 1978 9181' years
Fig. C.] - BRAZIL:PercentualDivergenceBetweenOfficialand AlternativeEstimatesof
EquilibriumExchangeRate
57. - 49 -
APPENDIX D
COMPUTATION OF FREE TRADE PRICE EQUIVALENT AT PARMGATE FOR
SELECTED PRODUCTS
(with the assistanceof Hugo Barros de Castro Faria)
The general procedure to obtain the farmgate free
trade price equivalent was the following:
1- obtain, for the products in question, its FOB
average export or import price (rice and wheat);
ii- subtract the transportation and other transaction
costs from this FOB price to obtain the price the
domestic producer would have received had his product
been sold at that FOB price.
In the work presented by CFP, the international
prices used (either for exports of imports) were obtained from
International Commodity Markets. We, here, use the average export
or import price received or paid along the year by Brazil. These
prices are registred by CACEX and do reflect better the international
market faced by Brazil than the alternative adopted by CFP.
In obtaining the free trade price equivalent, several
hypothesishad to be made. We based our hypothesis on the
58. - 50 -
previous work done by CFP (.1983).The products we consideredalso
were those considered by CFP as well as the items that comprised
the transaction costs to be subtractedfrom the FOB prices. The
computationsdone by CFP cover the period of 1977-83. In extending
this period back to 1966 we basically maintained the 1977-83
hypothesisby keeping the item as a percentage of FOB price or
by deflating the transaction cost item .The specific hypothesis
for each product is described in the footnotes to each product
table.
69. - 61 -
APPENDIX E
Free-Trade: Agricultural and Non-Agricultural
Value Added.
The value added for each agricultural product
was calculated according to the following procedure:
i) The only input considered was fertilizer,in
view of the difficulty in finding reliable price data for the
comparisons on international and domestic value added. Since
fertilizer is widely utilized in the crops considered, it does
not seem to distort very much the results to exclude other inputs;
ii) The technical coefficients were obtained
from Comissio de Financiamento da Producio (CFP) which has also
indicated the most commom formula for the components N,P and K;
iii) Domestic prices were obtained from Getulio
Vargas Foundation, Prices Paid by Farmers; international prices
were obtained from CACEX and are unit values;
iv) Credit subsidy was included in the
calculations. For the period 1966/1977 it was taken from Duran
(undated); for the rest of the period it was estimated by the
formula (1+ interest rate)/(l+inflation rate) where both the
numerator and denominator are semester rates. Notice that the
interest rates on loans to fertilizer purchases were zero from
1978 through 1981.
The value added for the non-agricultural sector
has been calculated as follows;
70. - 62 -
i) Estimates of effective protection for the
industrial sector were available for the years 1966, 1967, 1973,
1
1975 and 1981 . From theseestimates and the variation in the
import tariffs estimated in this work (see appendix B) we
calculated effective protection coefficients for the other years;
ii) From the National Accounts we obtained the
value added in the industrial sector which together with the
coefficients above allowed us to estimate the value added in the
industrial sector under free trade. To obtain the value added
in the non-agricultural sector under free trade we added to this
figure the value added in the service sector (the Income of the
Service Sector from the National Accounts).
The value added figures for the agricultural
products and the non-agricultural value added are presented in
the tables below.
All tables with an a .refer to the equilibrium
exchange rate E* while those without the a refer to Eb
1/ Bergsman-Malan (1970); Neuhaus-Lobato (1978) and Tyler (1983).
71. - 63 -
TABLE E.1 - BRAZIL: RICE - VALUE ADDED AND EFFECTIVEPROTECTION
DOMESTIC VALUE VALUE ADDED EFFECTIVE
YEAR -ADDED FREE TRADE PROTECTION
(VA RICE) (VA3 RICE) RICE
Cr$/Kilo Cr$/Kilo
1966
1967
1968
1969 0.290 0.382 - 24.238
1970 0.343 0.254 34.886
1971 0.526 0.271 93.837
1972 0.638 0.441 44.554
1973 0.649 0.875 - 25.874
1974 0.891 1.504 - 40.753
1975 1.599 2.140 - 25.296
1976 1.617 2.540 - 36.312
1977 1.672 3.235 - 48.331
1978 3.009 4.276 - 29.617
1979 5.509 7.437 - 25.932
1980 9.900 12.852 - 22.917
1981 16.468 22.963 - 28.283
1982 35.353 40.403 - 12.500
1983 98.588 90.679 8.722
Note: These calculationsdo not take into account the domesticprice
of triple superphosphatebefore 1973 since this informationis not
available. Before that year, we used the price of simple
superphosphate.
72. _ 64 -
TABLE E.la - BRAZIL RICE - VALUE ADDED AND EFFECTIVE
PROTECTION
DOMESTIC VALUE VALUE ADDED
YEAR ADDED FREE TRADE EFFECTIVEPROTECTION
(VA RICE) (VA3 RICE)
Cr$/Kilo Cr$/Kilo RICE
1966
1967
1968
1969 0.290 0.256 13.294
1970 0.343 0.163 110.475
1971 0.526 0.180 191.870
1972 0.638 0.291 119.287
1973 0.649 0.589 10.192
1974 0.891 1.231 - 27.590
1975 1.599 1.729 - 7.502
1976 1.617 1.948 - 16.946
1977 1.672 2.270 - 26.357
1978 3.009 3.073 - 2.075
1979 5.509 6.057 - 9.057
1980 9.900 12.943 - 23.511
1981 16.468 21.519 - 23.473
1982 35.353 41.525 - 14.862
1983 98.588 119.485 - 17.490
Note: These calculationsdo not take into account the domesticprice of
triple superphosphatebefore 1973 since this informationis not
available. Before that year, we used the price of simple
superphosphate.
84. - 76 -
TABLE E.8 - BRAZIL: Domestica / InternationalPrices at Equilibrium
Exchange Rate (Ewb)
YEAR NITROCALCIUM TRIPLE POTASSIUMCLORIDE
SUPERPHOSPHATEb
1966 0.898 0.713 1.647
1967 0.809 0.744 1.718
1968 0.874 0.761 1.556
1969 0.823 0.778 1.453
1970 0.867 0.802 1.344
1971 0.799 0.747 1.029
1972 0.889 0.611 1.267
1973 0.831 1.071 1.372
1974 1.298 1.093 1.893
1975 0.970 1.630 1.526
1976 0.686 2.148 1.528
1977 0.857 1.890 1.449
1978 0.841 1.722 1.398
1979 0.905 1.360 1.453
1980 0.885 1.629 1.498
1981 0.760 1.740 1.567
1982 0.680 2.399 2.039
1983 0.662 2.665 2.304
Note: a Domestic prices are averagesof the prices in the states of
Minas Gerais, Espitito Santos,Rio de Janeiro, Sao Paulo
Santa Catarina, Parana, Rio Grande do Sul, Mato Grosso, Mato
Grosso do Sul and Goias.
b Until 1972, simple superphosphate
85. - 77
TABLE E.8a. - BRAZIL: Domestica / International Prices at
Equilibrium Exchange Rate (Ek)
YEAR NITROCALCIUM TRIPLE POTASSIUMCLORIDE
SUPERPHOSPHATE
1966 1.331 1.056 2.442
1967 1.238 1.139 2.630
1968 1.223 1.065 2.177
1969 1.175 1.112 2.075
1970 1.247 1.154 1.934
1971 1.103 1.030 1.420
1972 1.253 0.861 1.786
1973 1.175 1.515 1.940
1974 1.539 1.296 2.245
1975 1.172 1.971 1.845
1976 0.874 2.739 1.949
1977 1.182 2.606 1.998
1978 1.133 2.320 1.884
1979 1.089 1.637 1.749
1980 0.880 1.619 1.489
1981 0.805 1.843 1.660
1982 0.663 2.340 1.989
1983 0.513 2.065 1.785
Note: a Domestic prices are averages of the prices in the states of Minas
Gerais, Esp!rito Santo, Rio de Janeiro, Sao Paulo, Santa Catarina,
Paran;, Rio Grande do Sul, Mato Grosso, Mato Grosso do Sul and
Goias.
b Until 1972, simple superphosphate.
86. - 78 -
APPENDIX F
EFFECTS OF AGRICULTURAL PRICE INTERVENTIONS ON
CONSUMPTION
The effects of agricultural price interventions on
consumption would be easy to compute in the absence of
commercialization costs and foreign trade. In this case, the
change in consumption could be calculated with the knowledgement
of the demand price elasticity for the product and the price
effect of the intervention computed at the producer level.
Unfortunatelyt this is not the case for the products
we are considering here. Not only commercialization costs
exist but distortions do exist in this activity. Thus, to
compute the price effect on consumption, due to price interventions
we propose to obtain the consumer price distortion-free as the
wholesale price plus the commercialization margin, both
evaluated in a distortion free situation. Knowing the actual
price paid by consumers and the corresponding distortion-free
consumer price, we can obtain the price distortion at the
consumer level (d) and therefore, compute the effects of price
intervention on consumption as follows:
d6 Qd = np . d (F1)
p
where
dQd = percentage change on domestic consumption due to price
intervention distortion d;
87. - 79 -
Np= own price elasticityof domestic demand;
PC - p
d = C where Pc is the actual price to consumers
C
and P* is the distortion-freeprice to consumers.C
To estimate P* we will need to estimate theC
commercializationmargin (cc) at wholesale level since the
distortion-freewholesale price (P*) can be estimated as weA
will demonstrate below and these two prices are connected
by that the margin as follows:
P* (1 + cc) = P* (F2)
A C (2
We propose the followingprocedure to obtain the
distortion-freeor the free-tradeequivalentwholesaleprices:
a) take the internationalprice for the good;
b) define this price in cruzeiros using the equilibrium
exchange rate (E*);
c) for imported goods, add to the internationalprice the
costs up to the wholesale level; for the exported goods,
substract from the exporting price the costs to take the
product from the wholesale level to the exportingport.
Proceeding in this fashion and using the data on the
Appendix D (tablesD) we obtained the wholesale price free-trade
equivalent as reproducedon tables F.1 and F.2 accordingwith
the equilibriumexchange rate used.
88. - 80 -
Having estimate PA, we depend on estimating cc to
obtain P*. Note that the distortions in the wholesale market
C
can be defined as follows.
PA (1 + d) PA (F3)
where PA is the actual wholesale price presented on table F.3.
Also, for the existing information, the observed
commercialization margin, that is, distortions included, can
be defined as;
PC = PA (1 + m) (F.4)
where Pc is the actual price paid by consumers as reported on
table F.4.
Thus, by definition, the observed commercialization
margin is the sum of the undistorted margin and the distortion:
PAm = PA cc + P* dcA A
or
cc = - m - dc (F.5)
A
Therefore, we can compute cc since m and dc are known:
m = -1 from (F.4) and PA and Pc are known since they are
p
observed prices (tables F.3 and F.4); and dc = P _ 1 from
A
(F.3) and PA is observed and PA has been estimated as described
aboveA
above.
89. - 81 -
As cc is computed by (F.5), the distortion free
consumer price can be obtained by (F.2). Since we have computed
two values for PA according with the equilibrium rate of
exchange was considered, we present on tables F.5 and F.6 the
corresponding values for P*. The corresponding price distortionsc
are presented on tables F.7 and F.8.
To compute the effects on the consumption according
to (F.1) the demand elasticities are required. We adopted the
following values:
Own Price Elasticity
Product of Demand
Corn - 0.90
Cotton - 0.55
Soybeans - 0.55
Rice - 0.18
Wheat - 0.60
These hypothesis were made based on the following
studies: corn: Nogueira - Brandt (1976) this elasticity refers to
the State of Sio Paulo;
Soybeans: Garcia (undated);
Rice: Crocomo (1982);
wheat: average value of -0.7 by Nogueira -Brandt (1976) and
- 0.544 by Garcia (undated).