the effects of income distribution on consumer imports

12
PHIUP ARESTIS Thames Polytechnic CIARAN DRIVER Thames Polytechnic and ,National Economic Development Office England The Effects of Income Distribution on Consumer Imports • The aim of this paper is to examine empirically the proposition that disaggregated income is important in determining the imports of consumer goods. An estimable equation is derived and tested using annual data (1962-84). The technique of Au- toregressive Generalized Least Squares is used throughout along with tests for sta- tistical bias. The results are supportive of the main hypothesis of the paper that income distribution exerts a powerful and statistically significant influence on the imports of consumer durables. Finally, the implications of income distribution on the balance of payments are discussed. 1. Introduction We have argued elsewhere that it is important to disaggregate income in a typical aggregate consumption function [Arestis and Driver (1980)]. This proposition is now extended to the case of the imports of consumer goods. The argument that the propensity to import consumer goods varies significantly by class of income is of interest in terms of eco- nomic theory and policy. If substantiated, it would imply that short- run consumer import models will have particular cyclical behavior and stability conditions. It would also suggest that policies or influ- ences that result in income redistribution could have a significant impact on the balance of payments. This short paper is intended to provide evidence that import propensities do indeed differ by income category. In what follows we first attempt to rationalize the proposition that imported consumer goods can be affected by income categories (Section 2). In Section 3 we put forward and discuss a relationship for imports of consumer goods that is explained by a number of *The views are those of the authors and not those of the institutions concerned. We are extremely grateful to two anonymous referees and to D.J. Smyth for helpful comments. We are, of course, completely responsible for any remaining errors, omissions and inelegancies. Journal of Macroeconomics, Winter 1987, Vol. 9, No. 1, pp. 83-94 Copyright © 1987 by Louisiana State University Press fll fld.flTf'kl/R7/.~ 1 .TO 83

Upload: philip-arestis

Post on 16-Sep-2016

218 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The effects of income distribution on consumer imports

PHIUP ARESTIS Thames Polytechnic

CIARAN DRIVER Thames Polytechnic and

,National Economic Development Office England

The Effects of Income Distribution on Consumer Imports •

The aim of this paper is to examine empirically the proposition that disaggregated income is important in determining the imports of consumer goods. An estimable equation is derived and tested using annual data (1962-84). The technique of Au- toregressive Generalized Least Squares is used throughout along with tests for sta- tistical bias. The results are supportive of the main hypothesis of the paper that income distribution exerts a powerful and statistically significant influence on the imports of consumer durables. Finally, the implications of income distribution on the balance of payments are discussed.

1. Introduction We have argued e lsewhere that it is impor tan t to disaggregate

income in a typical aggregate consumpt ion function [Arestis and Driver (1980)]. This proposit ion is now ex tended to the case of the imports of consumer goods.

The a rgument that the propens i ty to impor t consumer goods varies significantly by class of income is of in teres t in te rms of eco- nomic theory and policy. I f substantiated, it would imply that short- run consumer impor t models will have part icular cyclical behav ior and stability conditions. I t would also suggest that policies or influ- ences that result in income redistr ibut ion could have a significant impact on the balance of payments . This short pape r is in tended to provide evidence that impor t propensi t ies do indeed differ by income category.

In what follows we first a t t empt to rationalize the proposi t ion that impor ted consumer goods can be affected by income categories (Section 2). In Section 3 we put forward and discuss a relat ionship for imports of consumer goods that is explained by a n u m b e r of

*The views are those of the authors and not those of the institutions concerned. We are extremely grateful to two anonymous referees and to D.J. Smyth for helpful comments. We are, of course, completely responsible for any remaining errors, omissions and inelegancies.

Journal of Macroeconomics, Winter 1987, Vol. 9, No. 1, pp. 83-94 Copyright © 1987 by Louisiana State University Press fll f ld.f lTf 'kl /R7/.~ 1 .TO

83

Page 2: The effects of income distribution on consumer imports

Philip Arestis and Ciaran Driver

factors, including the "income distribution" variable. Section 4 dis- cusses the empirical results of the study, and Section 5 derives the implications of the empirical findings and puts together the main conclusions of the paper.

2. Income Distribution and Consumer Imports The argument that the propensity to import consumer goods

varies by class of income is directly analogous with the case that has been made for distinguishing income classes in consumer de- mand functions. There is a large literature on the latter question arguing for disaggregation of income either by functional category or by income bracket. One school draws on the views of classical economists [Kaldor (1956) and Pasinetti (1962)]. The argument can also be posed in terms of the marginal utility of money as well as differing variability of types of income. The greater variation in in- comes of the non-wage category means that a larger cushion of as- sets is required than with the wage category where incomes are more stable. 1 This argument is reinforced if the non-wage category

cludes economic units who are risk averters and if their future owledge of capital markets is not perfect [Burmeister and Taub-

man (1969)]. Furthermore, the influence of income distribution on consumer demand in general and on demand for consumer ira- portables can be justified within the framework of Muellbauer (1976). In an attempt to specify the requirements for community prefer- ences, he derives nonlinear Engel curves that imply a behavioral role for income distribution.

Quite apart from the distribution questions that influence con- sumption generally and consumer importables as a consequence, there is a further reason why imports of consumer goods may be influ- enced by income distribution. This concerns the composition of im- ports in consumer expenditure, in terms of the characteristics of the goods via-~-vis domestically produced consumer goods and ser- vices. In development theory, for instance, the low income levels of wage workers that restrict their consumption to basic needs, gen- erally met by local production, have often been used to support the argument that balance of payments difficulties are due to ex- cessive imports of luxuries for the managerial or land-owning classes.

1Arestis and Driver (1980)--see also the references therein--provide evidence to suggest that the marginal propensity to save out of unearned and self-employed income is greater than that out of wage income.

84

Page 3: The effects of income distribution on consumer imports

Income Distribution and Consumer Imports

But in industrialized (or postindustrial) economies, such as the U.K., no such rationale applies. In these economies those in rece ip t of wages and salaries as a p r ime source of income may have a g rea te r propensi ty to spend their marginal income on impor t ed commodi - ties, such as cars and consumer durables, than do the recipients of unearned income and the self-employed. The lat ter may be ex- pected to spend more of their marginal income on such i tems as land, second homes, house improvements , art objects, private health and education or o ther services, all of which mos t certainly have a low impor t content. 2

3. The Model As our starting point we employ the basic identity:

--r (NQC)t L( g),-iJ (1)

where (NQC) is nominal imports of consumer goods, (QC) is real imports of consumer goods, (PQ) is the pr ice of consumer impor ts and t is a t ime subscript. Fu r the rmore , the ratio [(QC)t/(QC),-1] can be thought of as a "vo lume revision" variable in that it is con- sistent with the following formulation.

Impor t s of consumer goods in real t e rms (QC) can be assumed to be de te rmined by two sets of decisions: by the p lanned vo lume of QC (QC v) and the vo lume that is revised (QC R) within the t ime periods from the plan to the cur rent date. W e may thus write:

QC = (QC e + QCR), (ia)

or in nominal terms:

PQCQC) = (QC e + Qca)PQ,

ZThis hypothesis receives some support from statistics on the share of services in consumption by income bracket. Figures for household consumption show that for those on above-average incomes the proportion of income spent on services more than doubles with a doubling of income [Gershuny (1983)]. If there is a strong correlation between non-wage incomes and higher incomes and also a strong cor- relation between services and non-traded activities, these figures provide prima fa- cie support for the hypothesis that consumer imports are more affected by a rise in wage income than by non-wage income.

85

Page 4: The effects of income distribution on consumer imports

Philip Arestis and Ciaran Driver

or

X Q C = (QC e + QcR)PQ. (lb)

Furthermore, it can be assumed that QC e is generated by an adaptive expectations process so that:

p c ~, = QC,_t + k[OC,_~ - fCet_l],

Since by definition

QC R= QC - QC e ,

0 < k ~ 1. (lc)

(ld)

it follows that by substituting (lc) and (ld) into (lb) we arrive at:

NQCt = [kQCt-~ + (QC - hQCt-1)]PQ, (le)

or

NQCt = [1 + QCt/kQCt-I - 1]hQct_~[PQt_i/eQ,-t]PQ, (if)

or

NQC, = [QCJQC,_I][PQ/PQ,_I]NCQ,_I,

which is, of course, identity (1). Identity (1) can be transformed into a behavioral equation by

modelling the determinants of the first term in square brackets• In Equation (2), below, [(OC)t/(QC)t-1] is proxied by a geometrically declining distributed lag in income distribution (YDw/YDc), income (YD), visible imports (IVVQ), relative prices (PQ/PC) and capacity utilization (CU), in a multiplicative fashion. Thus we have:

[ (QC)t 1 = m [(yDw/YDc)t_i+l la~.l_[" [(YD)t_j+!ld2

L ~ J ao ~ t (rDw/Yoo),_, J ~:~ L (Yo),_j J

r"

• ( l hffil L (IWQ)t-h -I ~=1 L (PQ/PC)t-+ J

+ [ ]°~ • [ - [ (cv),_o+l o=~ L (cu),_o j

(e)

86

Page 5: The effects of income distribution on consumer imports

Income Distribution and Consumer Imports

The exponential character of this relat ionship has some inter- esting advantages: it enables one to obtain the elasticities straight- forwardly, and, p e r h a p s more important ly , this way of formulat ing volume revisions implies that they react propor t iona te ly to a rise and a fall in the variables that influence t h e m [Barker (1970) p. 126]. We have assumed that in (2), in addit ion to real disposable income (YD), a distribution variable (YDw/YDc) also affects vo lume revision. One of the central proposit ions of this p a p e r is that, even if disposable income were to be the same, vo lume revision can still change ff there are different propensi t ies to impor t consumer goods. The income-distr ibut ion ratio (YDJYDc) purpor ts to capture this influence. YD~ refers to wages and salaries plus pay of H M forces net of income tax, surtax and employees ' national insurance con- tributions; and YDc refers to income from rent , dividends, i n t e r e s t and se l f -employment net of tax and national insurance contr ibut ions of self-employed and nonemployed persons. YD is real personal dis- posable income which consists of the sum of YD~ and YDc plus YDa where the latter is the residual in the ident i ty YDR = YD - YD~ " YDc. The variable YDa is ignored in what follows s ince it is a residual category. ~,

The 1VVQ variable represents an index of the vo lume of vis- ible imports. I t is in tended to capture variation in all s tructural variables which affect impor ts s and it may also capture the effect of some of the cylical changes in impor ts which are somet imes ac- counted for by changes in capacity utilization or o the r measures in economie activity. However , a capacity utilization variable is in- cluded separately, 4 al though it could be expec ted already to be proxied by the index of visibles. W e r e it not included, there would be a legit imate doubt as to whe the r the distr ibution variable was

3Among these structural variables are long-run trend influences reflecting, in part, deindustrizliT~tJon and structural changes in the world economy [Blackaby (1979) and Parboni (1981)]. The IVVQ variable has also been employed [Beenstock and Warburton (1980)] to reflect any changes caused by the phasing out of various tariffs as the U.K. made a transition to free trade.

4Thirlwall (1980) has argued that the appearance of a capacity variable in an import equation amounts to saying that the domestic price level does not fully ad- just to balance domestic supply and demand: imports fulfill the balancing role. In an analysis of import penetration in the same work Thirwall has shown that, at least in some industries, there is a cyclical as well as a trend effect. The cyclical effect is associated with shortages in skilled labor and with other cyclical indicators, such as capacity utilization [see also (Akhtar (1981)]. The importance of a cyclical variable can readily be accepted while noting that imports have risen sharply even in years of spare capacity [Panic (1975)].

87

Page 6: The effects of income distribution on consumer imports

Philip Arestis and Ciaran Driver

capturing some of the cyclical variation normally attributed to the capacity term. The capacity utilization variable used in this study is constructed as one minus the annual average of the percentage of those working at less than full capacity as reported in the CBI Industrial Trends Survey. We have employed a measure of capacity utilization in the consumer goods industries for the years that it was available (1967-84) and an index for total manufacturing for the ear- lier years (1962-66). These two measures are highly correlated throughout the period and differ by only one percentage point in the changeover year, 1967. (The capacity utilization index for total manufacturing for the entire period was also tried and produced very little change in the results.) A "'relative price" variable, de- fined as PQ (the price deflator of imported consumer goods) over PC (the price deflator for consumer goods), is also included, s In view of the large exchange rate changes in recent years and in view of the fact that our study extends to the mid-1980s we should ex- pect a significant "relative price" effect.

We may now substitute Equation (2) in (1) and upon taking logs we arrive at:

a(Inpc)~ = a ( InNpC) , - a(IneO)~ = C N S r + ~ a[a In ,~1 L YDcJt-i+l

r l r

+ 2 a~A(InYD)t_j+I + 2 ahA(InlWQ)t-h+l j = l h : l

• I P Q 1 + ~] a]A In + a~A(InCU),_~.l + ut k=~ L ecJ, -k+l o°1

SThe ratio (PQ/PC) is the usual way of modelling relative price effects, although tariff changes should appear as a separate variable [Morgan (1975) and Humphrey (1976)]--but see Whitley (1979) who fails to establish a separate tariff effect. All these indices suffer from some major weakness, particularly that they do not fully provide for quality change over time and that the possible endogenous nature of import prices is not accounted for. Perhaps not surprisingly, earlier studies have found little role for a "relative price" effect at an aggregate level [Barker (1970)]. It may very well be, however, that these early unsatisfactory results are due to the argument advanced by Learner (1981), that in reality what one estimates is an "at- tenuated supply function." One should note that price effects appear to be more significant at a disaggregated level [Hotson and Gardiner (1983) and see also Haynes and Stone (1983)].

88

Page 7: The effects of income distribution on consumer imports

Income Distribution and Consumer Imports

where ut is the error term and i, j, h, k and v are lags to be de- termined empirically.

4. Empirical Results 6 When annual U.K. data was collected for the per iod 1962-84,

the results repor ted in Table 1 were obtained. Annual data is used throughout because YDw and YDc are available on an annual basis only. The 1962-84 per iod is dictated by the availability of data on the QC variable. The latter refers to imports of consumer goods 7 (the variables YDw, YDc and YD as well as PQ, PC and CU are defined above; IVVQ refers to imports of visibles by volume). It must be stressed at the outset that the dependen t variable is QC and not NQC. The reason for constraining the coefficient on PQ to be unity is that it makes both theoretical sense and enjoys empirical verification too. Early attempts to estimate NQC relationships clearly indicated to us the validity of the unit coefficient p roper ty of the PQ variable. 8

In terms of the diagnostics we have cited on Table 1, the fol- lowing commerits are in order. The technique of Generalized Least Squares has been used throughout with the assumption of first- order autoregression being adopted. 9 In o ther words the scheme. ut = put-t + ~t is employed in all est imated equations where p is the autocorrelation coefficient. R z and D W are the standard coef- ficient of determinat ion and Durbin-Watson statistics respectively, with S being the standard error of the regression. ×~(n) is a like- lihood ratio test which is asymptotically a chi-square test of the va- lidity of the autoregressive restrictions (where n is the n u m b e r of restrictions). The hypothesis here is that the autoregressive restric- tions are correct, and therefore if x~(n) is greater than the critical value we reject the restriction which suggests reformulat ing the dy- namics of the model. This particular statistic is, therefore, effec-

SAil the data was collected from National Income and Expenditure (various is- sues) except for (YD) and (IVVQ), which were collected from Economic Trends (various issues). Also, where appropriate, all the data is at constant 1975 prices.

rWe are extremely grateful to Mr. Kirby of the Department of Trade who very -kindly provided some data for the QC variable. Also to the CBI for providing fig- ures for the capacity utilization variable for some of the earlier years.

8We are mindful of the proposition by Leamer and Stern (1970) that if the im- port price deflator is not reliable, imports should be specified in money terms.

~The computing program we have utilized is called GIVE and has been written by Professor D. Hendry of Oxford University [see Hendry and Srba (1978)].

89

Page 8: The effects of income distribution on consumer imports

Philip Arestis and Ciaran Driver

TABLE 1. Estimated QC-Equations. Estimation Period: 1962-1984 (t statistics in brackets)

Dep. Vars. :~ (1) (2) (3) (4) ~ Ind. Vars. and zl(InQC)t a(InQC)t a(InQC), A(InQC)t

Diagnostics

CONSTANT 0.005 0.010 0.012 0.016 (0.25) (0.42.) (0.52) (1.64)

AIn(YDw/YDc)t 0.590 0.600 0.572 0.583 (2.24) (2.22) (2.16) (2.14)

AIn(YDoJYDc)t_ 1 - - - - O. 209 0.203 (0.90) (0.86)

A(In YD)t 1.283 1.348 1.285 1.344 (Z. 02) (2.04) (2.01) (2.01)

AIn(PQ/PC)t_I -0 .636 -0 .675 -0.531 -0 .569 (2.77) (2.75) (2.05) (2.06)

A(In IVVQ)t 1.531 1.398 1.442 1.322 (4.21) (3.11) (3.81) (Z. 87)

A(In CU)t - - 0.043 - - 0.039 (1.53) (0.48)

p 0. 372 0. 423 0. 340 0. 401 (1.52) (1.75) (1.31) (1.57)

R 2 0.72 0.72 0.73 0.73 D.W. 1.32 1.28 1.49 1.43 S 0.07 0.07 0.07 0.07 X~(n) 1.44(4) 2.64(5) 6.00(4) 6.76(5) X~(1) 0.80 1.02 1.37 1.54

tively a test of the dynamic specification of the estimated equation [for more details on the econometric methodology pursued, see Hendry (1976) and Appendix 5 in Arestis, Frowen and Karakitsos (1978)]. Finally, concerning diagnostics , there is the x~(n)-statistic which is the Engle's (1982) ARCH test (autoregressive conditional heteroscedasticity), which tests the hypothesis gl = 0 in the model:

90

Page 9: The effects of income distribution on consumer imports

Income Distribution and Consumer Imports

E[u~/u,_l ] = go + glu,~-~.

This diagnostic is asymptotically distributed as ×2(1)--with one de- gree of freedom.

It is clear from Table 1 that, in addition to the usual income effects, there is also evidence of distributional effects. In all cases reported in Table 1 the coefficient of the income distribution vari- able is remarkably stable in that it is hardly affected by different specifications. Nor is it affected by differing lag structures in one or more variables that belong to the set of explanatory variables. The coefficients are around the 0.5 to 0.6 mark with respeet to the contemporaneous ratio (YDw/YDc). The lagged (YDw/YDc) variable is insignificant and can, therefore, be dropped from the A(InQC) equation without any loss of the significance of the remaining vari- ables or worsening of the diagnostics. There appears, therefore, to be strong evidence that when the rate of growth in wages and sala- ries rises relative to the rate of growth in other incomes the rate of growth in imports also tends to rise.

In Table 1 we report four equations; in two of them the ca- pacity utilization term is omitted. The inclusion of the CU variable produces very little, if any, overall change in the results, with t he variable itself being insignificant. In Equations (2) and (4) the ca- pacity variable increases the values of the two X z statistics. Clearly, the dynamics of the equations which exclude capacity utilization are more satisfactory than those equations which include it. For this reason CU has been retained in the regressor set, even though the t values are insignificant. These results, therefore, do not seem to support the argument that because of the interdependence of mea- sures of economic activity and capacity utilization, omissions of any one of them could lead to apparently significant results for the in- cluded variables [Barker (1979)]. It would seem to be the case that our income variables measure pressures of demand and distribu- tion, while the capacity utilization variable measures the extent of excess demand due to shortages of domestic supply [see Akhtar (1981)], so that the degree of interdependence may not be danger- ously high. One should also note the significance of the coefficient in imports of visibles which, to repeat, stands for a number of other quantifiable and nonquantifiable variables. It would appear that the performance of the relative price variable lagged once is quite sat- isfactory; it is properly signed and significant and also stable. This is an important further finding given that the results of U.K. studies on this particular variable are not always very successful. It is u s u -

91

Page 10: The effects of income distribution on consumer imports

Philip Arestis and Ciaran Driver

ally the long-term coefficient that turns out to be negative and sig- nificant [Akhtar (1981)]. In the latter study it is, in fact, suggested that the adjustment lags with respect to prices are as long as over three years. In our present study the lag is clearly shorter.

In estimating our equations, we must recognize the possibility of statistical bias. Such biases could result from omitted variables and simultaneity or an incorrect functional form. A simple proce- dure to test for the presence of such biases has been designed by Ramsey (1969) and modified by Ramsey and Schmidt (1976). It in- volves a three-step procedure: (i) run the original regression and obtain ~, the values of the fitted dependent variable, (ii) rerun the regression with ~z, ~3 and ~4 included as additional regressors, and (iii) the test is an F-test of the joint significance of the ~.z, ~3, and ~4 variables. An insignificant F indicates that whether or not there are theoretical reasons for suspecting problems due to omitted vari- ables or simultaneity, the resulting biases in the coefficients of the included variables are statistically insignificant. This test was carried out on all four equations in Table 1. The resulting F statistics were respectively: F3,14 = 0.733; F3.13 = 0.813; F~.13 = 0.474 and F3.12 = 0.410. These are all w~] below the corresponding critical values at the 5% level.

5. Conclusions and Implications The main conclusions to emerge from our study are clear:

(i) Income distribution exerts a powerful and statistically sig- nificant influence on the imports of consumer durables when wages and salaries rise relative to other incomes imports rise. This is in accord with the arguments advanced earlier in this regard.

(ii) Visible imports seem to perform well as a proxy for trend changes, tariffs and other shocks. The total disposable in- come variable also performs as expected.

(iii) The "relative price" variable seems capable of improving the performance of the equation (this is less so in the case of the "capacity utilization" variable).

The policy implication for the U.K. centers around the first conclusion: policymakers with a preference for redistributing in- come towards those in receipt of wages and salaries and away from

92

Page 11: The effects of income distribution on consumer imports

Income Distribution and Consumer Imports

those in receipt of unearned income must consider very carefully the short-term impact of such a move on the balance of payments.

Received: October 1984 Final version: June 1986

References Akhtar, M.A. "Income and Price Elasticities of Non-Oil Imports for

Six Industrial Countries." The Manchester School of Economic and Social Studies 49 (December 1981): 334-47.

Arestis, P., S.F. Frowen and E. Karakitsos. "The Dynamic Impacts of Government Expenditure and the Monetary Base on Aggre- gate Income: The Case of Four O.E.C.D. Countries, 1965-1974." Public Finance 33, nos. 1 and 2 (1978): 1-21.

Arestis, P., and C. Driver. "Consumption Out of Different Types of Income in the U.K." Bulletin of Economic Research 32 (No- vember 1980): 85-96.

Barker, T.S. "Aggregation Error and Estimates of the UK Import Demand Function.'" In The Econometric Study of the United Kingdom. K. Hilton and D.F. Heathfield, eds. London: Mac- millan, 1970: 115-145.

Barker, T.S. "Identification of Activity Effects, Trends and Cycles in Import Demand." Bulletin of the Oxford Institute of Eco- nomics and Statistics 41 (February 1979): 63-68.

Beenstock, M., and P. Warburton. "'UK Imports and the Interna- tional Trading Order." Discussion Paper No. 83. Economic Fore- casting Unit, London Business School: July 1980.

Blackaby, F., ed. Deindustrialization. London: Heinemann, 1979. Burmeister, E., and P. Taubman. "'Labour and Non-Labour Income

Saving Propensities." Canadian Journal of Economics 2 (Feb- rnary 1969): 78-89.

Engle, R.F. "Autoregressive Conditional Heteroscedastieity with Estimates of the Variance of United Kingdom Inflations." Econ- ometrica 50 (July 1982): 987-1008.

Gershuny, J. Social Innovation and the Division of Labour. Oxford: Oxford University Press, 1983.

Haynes, S.E., and J.A. Stone. "Specification of Supply Behaviour in International Trade." Review of Economics and Statistics 65 (November 1983): 626-32.

93

Page 12: The effects of income distribution on consumer imports

Philip Arestis and Ciaran Driver

Hendry, D.F. "'The Structure of Simultaneous Equations Esti- mators." Journal of Econometrics 4, no. 1 (1976): 51-88.

Hendry, D.F., and F. Srba. Technical Manual for GIVE. London School of Economics, 1978.

Hotson, A.C. and K.L. Gardiner. "'Trade in Manufactures." Dis- cussion Paper No.5 (Technical Series). Bank of England: Decem- ber 1983.

Humphrey, D.H. "Disaggregated Import Functions for the UK, West Germany and France." Bulletin o f the Oxford Institute of Eco- nomics and Statistics 38 (November 1976): 281-98.

Kaldor, N. "Alternative Theories of Distribution." Review of Eco- nomic Studies 23, no. 2 (1956): 83-100.

Leamer, E. "'Is it a Demand Curve or is it a Supply Curve? Partial Identification Through Inequality Constraint." Review of Eco- nomics and Statistics 63 (August 1981): 319-27.

Learner, E., and R.A. Stern. Quantitative International Economics. Boston: Allyn and Bacon, 1970.

Morgan, A.D. "Tariff Reductions and UK Imports of Manufactures: 1955-71." National Institute Economic Review 75 (May 1975): 38- 54.

Muellbauer, J. "'Community Preferences and the Representative Consumer." Econometrica 44 (September 1976): 979-99.

Panic, M. "Why the UK's Propensity to Import is High." Lloyds Bank Review 115 (January 1975): 1-12.

Parboni, R. The Dollar and its Rivals. London: New Left Books, 1981.

Pasinetti, L.L. "Rate of Profit and Income Distribution in Relation to the Rate of Economic Growth." Review of Economic Studies 29 (October 1962): 267-79.

Ramsey, J.B. '~l'ests for Specification Errors in Classical Linear Least- Squares Regression Analysis." Journal of the Royal Statistical So- ciety 31, series B, no. 2 (1969): 350-71.

Ramsey, J.B., and P.C. Schmitt. "Some Further Results on the Use of OLS and BLUS Residuals in Specification Error Tests." Jour- nal of the American Statistical Association 71 (June 1976): 389- 90.

Thirlwall, A.P. Balance-of-Payments Theory and the United King- dom Experience. London: Macmillan, 1980.

Whitley, J.D. "Imports of Finished Manufactures: The Effects of Prices, Demand and Capacity." The Manchester School 47 (De- cember 1979): 325-48.

94