Journal
of Public Economics
1 (1972) 159-167.
0 North-Holland
Publishing
Company
BOOK REVIEWS
G. Fromm Amsterdam
ted.), Tax incentives and capital spending (Washington, and London, North-Holland. 1971, xxvii + 301 pp.)
Brookings
Institution,
This book contains four long essays, two comments, and an editorial introduction, all assessing the effects of the investment tax credit and accelerated depreciation applicable to fixed investment expenditure by United States corporations. The papers and comments were presented to a Brookings conference of experts held in Washington in November 1967, but the volume did not appear until mid-197 1, As a record of important investment studies, this book is an outstanding contribution. The conference organizers were both wise and fortunate to persuade energetic and talented scholars of investment behaviour to use their latest equations to help assess the effects of tax incentives. The papers themselves are much more than conference pot-boilers for they contain the results of new efforts to specify plausible roles for tax variables in the definition of desired capital stock and in the timing of the resulting investment expenditure. Some of the major theoretical and empirical results (notably those of Bischoff) have existed for some while but are published here for the first time. From the viewpoint of the economist concerned primarily with assessing the results of public policies, however, the book is disappointingly difficult to use. The problem with the conference studies as a joint guide to the influence of tax incentives is that there are too many unnecessary and unanalyzed differences in data, definitions of variables, model structure, and presentation of results. The editor and the discussants (Fisher and Harberger) alike report that the four studies produce very different assessments of the effects of tax incentives on investment, but they and the interested reader are hard put to say why and by how much the answers differ. This situation could have been largely avoided if the paper-givers had all been asked to estimate the effects of (at least) a number of precisely defined tax incentives, all measurements to be made under specified initial conditions. These requirements would have meant little extra work for the researchers, and would have rendered their results far more useful. More effort would have been required to achieve a uniform level of aggregation, and to assess separately the effects of some of the important differences between the models used, but both these developments are necessary in order to achieve the full gains to be made from bringing independent approaches to bear on the same subject. Some examples may help to demonstrate this point as well as to preview the book for potential readers. A six-way split of business investment helps to classify the studies. Expenditure on equipment can be divided by industry: manufacturing (ME), regulated industries (RE) and other industries (OE). A parallel split-can be made for investment in structures (MS, RS, and OS). The following table classifies the-four studies by the number and nature of the investment equations, data used, and the methods employed to estimate the effect of tax incentives. Differences in aggregation vitiate simple comparisons of the equations because tax incentives and investment behaviour alike differ by industry and by type of asset. The investment tax credit applied only to equipment, and to a greater extent in manufacturing than in other
160
Authors
Book reviews
Fquations estimated
Data
Hall and Jorgenson
ME,RE+OE MS,RS+OS
Bischoff
k structure
Estimation method
Assessment method
Annual 1935-40 and 1954-65
Polynomial, same for output and relative prices
OLS with correction for autocorrelation
Single equation
ME+RE+OE
Quarterly 3Q19514Q1965
Polynomial, different for output and relative prices
Non-linear
Single equation
Coen
ME+MS
Quarterly lQ1950341966
Inverted-V plus Koyck transform
OLS
Single equation
Klein and Taubman
ME+MS RE+RS OE+OS
Quarterly
a priori polynomial plus Koyck transform
OLS, with and without 2nd order autoregressive transformation
Simulations using Wharton model
OLS = ordinary
least squares.
industries. Similarly, the accelerated depreciation measures assessed have effects that differ by type of asset. Further differences among the various results can be traced to theoretical assumptions and estimation methods for the investment equations, and to assumptions about what is happening in the rest of the economy. I shall mention only those differences most likely to lead the four papers to different conclusions. Hall and Jorgenson assume that factor proportions are equally variable ex ante and ex post, and constrain the desired capital stock to have the same elasticity and time pattern of response with respect to changes in output and in relative prices. Bischoff makes the more reasonable assumption that factor substitution is easier ex ante than ex post. His separate lag distributions for output and for relative prices indicate much slower response to changes in relative prices than to changes in output, although the long-term elasticity of desired capital is about 1.0 with respect to both output and relative prices. Coen’s model differs from the others by allowing cash flow to affect the adjustment speed in a way, however, that is plausible only if yet investment is positive. This provides another channel of influence for accelerated depreciation and other tax incentives that alter cash flow. In contrast to Hall and Jorgenson and Bischoff, Coen uses the price of capital services divided by the wage rate, rather than the price of output divided by the price of capital services, as his relative price variable. Like Bischoff, Coen estimates the effect of relative prices separately from that of output. Coen’s estimated elasticity of desired capital with respect to the price of capital services, the main variable through which the tax incentives operate, ranges between -.05 and -.58, This may be because, as Harberger and Griliches have suggested, the measurement error is greater for the cost of capital services than for the other determinants of the desired capital stock. On the other hand, the use of relative costs, rather than the price of output relative to price of capital services, leads us to expect an elasticity of desired capital with respect to the
Book reviews
161
price of capital services of less (in absolute value) than the elasticity with respect to expected output, even within a strict neo-classical framework. Assuming a two-factor Cobb/Douglas production function, the elasticity of desired capital with respect to the price of capital services is -1.0, if the partial condition for profit maximization is used, and equal to the ratio of the labour exponent to the sum of the labour and capital exponents (in the production function), if the cost minimization condition is employed. In a world of constant returns and perfect competition, both approaches would yield the same result for the desired capital stock, because of the macro relation between input prices and output price. In that world, or in any more realistic world, where output price is influenced by the price of capital services, it is a mistake to assess the influence of tax incentives without taking induced changes in output price into account. The Klein and Taubman study is the only one taking account of how investment is influenced by induced changes in prices, output, interest rates, and other macro variables indirectly affected by tax incentives. There was some feeling at the conference, as reported in the discussion, that the Wharton investment equations were inadequately specified (as linear functions of output, cash flow, long-term interest rates, and the lagged capital stock) to make the best use of Klein and Taubman’s careful analysis of specifix tax measures and the simulation possibilities provided by the Wharton model. Their macroeconomic approach, which they supplemented by direct questionnaire evidence, will no doubt have to be followed by others wishing to assess the advantages and disadvantages of tax incentives for investment. Klein and Taubman assume single-equation investment effects in 1967 averaging .9 billion dollars (p. 233) resulting from a suspension (4Q6664467) of the investment tax credit and accelerated depreciation. In complete model simulations, the induced 1967 changes in investment totalled 1.6 billion dollars, larger than the single equation effects, presumably because of the accelerator influences flowing from the induced changes in output. The induced changes in GNP totalled 2.8 billion dollars. Harberger’s comments suggest that a model more supply-oriented than the Wharton model would indicate more substitution between categories of investment and smaller changes in GNP. If, for example, real output were subject to a strict ceiling, then any taxinduced increase in plant and equipment investment would lead to some mix of increased net imports, increased prices and wages, and decreased residential construction and expenditure by governments and households. Under these circumstances, the net changes in business investment, as predicted by model simulations, could well be less’than those estimated from the investment equations on their own. These issues, scarcely mentioned in the present volume, lead to further questions of central importance to public economists assessing the impact of tax incentives on the economy. Do tax incentives make it easier to reconcile domestic employment and balance of payments objectives? How do their net effects depend on the underlying balance between supply and demand, and on the choice of companion government expenditure and financial policies? In summary, the book will prove frustrating to those who wish either to make quantitiative comparisons among the various equations and results, or to understand the macroeconomic consequences of investment tax incentives. Nevertheless, the book contains some of the most interesting investment studies of the 1960’s and is required reading for all those who wish to go further in the quantitative study of tax incentives for investment.
University
John Helhwell of British Columbia