Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Appendix A Estimates of Comparative Productivity and Costs UnderA~ternative Methods This appendix presents estimates of the U.S.-Japanese landed-cost differential on a small vehicle, using several different methods and approaches. In addition, estimates of differences in produc- tivity also were obtained. Some of the panel members have had access to internal studies of these issues that make use of pro- prietary data. Our intent here, however, is to make use of public information in order to illustrate the range of estimates implicit in generally available data. The four approaches can be distinguished in their unit of analyses. The first adopts what is essentially an economy-wide m acro perspective; this we have labeled the industry/macro approach. The second, the industry/micro view, looks at the issue from the standpoint of the industry taken as a whole. The third, the company perspective, uses data on two major firms, as detailed in their annual reports. Finally, we present data on plant_by-plant comparisons of particular kinds of production processes. The concluding section of this appendix summarizes the results. Before reviewing the analysis it is important to note the difficulties associated with calculations of this sort. The auto industries of the United States and Japan produce a different mix of products and have organized production in different ways, particularly in terms of vertical integration. Productivity com- parisons are also significantly affected by differences in capacity utilization that have been substantial in recent years. While attempts have been made to correct for these factors, even the most careful comparison requires judgements and assumptions that affect the results. The Industry/Macro Approach One way to compare productivity in auto manufacturing in the United States and Japan is to do the comparison on an economy- 169
170 wide basis. This approach, which involves comparisons of effici- ency in every production activity required in making an automo- bile (i.e., mining iron ore, oil refining, steelmaking, machinery fabrication, power generation, business services), uses an input- output analysis. Given the scope of automobile production, this is close to comparing the economy-wide productivity of the United States with that of Japan. Available data on economy-wide productivity indicate a U.S. advantage, so it is no surprise that the industry/macro approach tends to find higher total labor hours per vehicle (defined over the whole economy) in Japan. Dan Luria of the United Auto Workers' Research Staff has used 1977 input-output data updated to 1980 and estimated labor content for a small, 1980 Japanese vehicle to be 364 hours versus 336 for the American; this is a U.S. advantage of about 8 percent. With assumed hourly charges of $ 11 (averaged over all embodied hours) for the United States and $7 for Japan, Luria's estimate of the manufacturing cost advantage of the Japanese is $1148. If freight charges ($400) are subtracted, Luria's analysis implies a landed-cost advantage of $748. It should be noted that, while the industry/macro analysis does not provide an estimate of productivity or cost difference origi- nating within the auto sector, it is not inconsistent with the existence of a U.S. disadvantage. Indeed, existing evidence suggests that a good part of the production activity outside of the auto sector (and even some in the sector but outside of the big firms) may take place in small establishments where productivity is low. Luria suggests that the Japanese had a productivity advantage of 11 percent (about 15 percent in 1981 terms) in the auto sector but that they were 16 percent less productive in other industries. Industry/Micro Costs and Productivity: Motor Vehicle and Parts Industry To obtain estimates of cost differences in the auto industry itself, it is useful to first examine differences in labor productivity. In 1974 Baranson estimated that output per labor hour in the Japanese motor vehicles and parts industry was 88 percent of the level reached in the United States (i.e., the ratio of productivity in Japan to productivity in the United States was 0.88~.2 This 12 percent U.S. advantage in 1974 is consistent with calculations developed by the British Central Policy Review Staff; using 1973 data they estimated the relative productivity ratio to be 0.82.3 These estimates can be updated using published data on growth rates of productivity in the United States and Japan. Abstracting from cyclical fluctuations, the evidence suggests that growth in
171 labor productivity in the Japanese auto industry (vehicles and parts) averaged 8-9 percent per year in the 1970s; the comparable figure for the United States is 3-4 percent.4 If we apply a mid-range estimate of the differential (i.e., 5 percent) to previous estimates of relative productivity, we arrive at a value for 1980 of 1.18. The data on industry growth rates imply that the Japanese producers operated at levels of productivity almost 20 percent above their American competitors. More rapid growth in productivity in Japan has been accom- panied by higher rates of wage increase. Whereas in 1974 Japanese hourly compensation rates were about 37 percent of the U.S. figure, in 1980 the ratio was roughly 50 percent.5 Relative unit labor costs can be calculated by dividing the compensation ratio by the index of relative productivity.6 This method yields an estimated unit labor cost ratio of 0.424 = (0.5/1.18) for 1980. Thus, Japanese-U.S. differences in the growth of productivity and compensation have been offsetting; the wage gap has continued to narrow, while a productivity gap has emerged and grown larger; relative unit labor costs remained roughly constant at 0.425 over the 1974-1980 period. We assume that the estimate of relative unit labor costs applies both to the auto manufacturers and to their suppliers. The estimate can be used to calculate the dollar value of the Japanese advantage in lower overall labor costs per vehicle. Since our focus is on the overall labor content in the vehicle, including purchased parts and materials, we are in effect comparing the relative position of the Japanese and U.S. automobile production systems. We are not comparing labor costs at the level of a Toyota or a General Motors' (GM) car; our analysis seeks to estimate the impact of productivity and compensation in the whole productive confederation--original equipment manufacturers (hereafter, OEMs) and suppliers of components and materials. Note that this is different from the industry/macro approach, which included many more activities outside the "productive confederation" in the calculations. Table A. 1 presents the basic building blocks of the analysis. Column 1 contains an estimate of the shares in manufacturing costs of hourly and salaried labor (at the OEM level), purchased components, and materials. These estimates are based on data prepared for the National Research Council, Committee on Motor Vehicle Emissions, as well as discussions with industry sources.7 The estimates do not reflect the experience of any one company but are intended to approximate an industry average. It should be emphasized that all of the data refer to production of a small, subcompact vehicle. In addition to the data in column 1, industry participants have provided us with estimates of average OEM labor hours per vehicle, current rates of employee cost per hour,
172 - Ct ~ Cal o ~ ~ Ce X CQ ~ O ~ ^ ~ ~ ~ _ - , .o Cal Cal o V) so clot o V so a' _) V: o o Cal Cal EM - , of _ oo o cr _ - Oo ~ ~ x ^ o ~ ~ _ _ ID ~ o At. ~~ ~ ~ ~ ~ ~ as_ , _ _ an 04 0 a ~ ~ 0 0 ~ ~ ~ ~ ° ~ ~ O O :^ ~ u, ~ - o ~ ~ . oo ~ ~ c~ u, c-) a~ ~ ,,= ~o a~ .= o ~ a~ ~t ~ ~ - ~ ~ ~ - s~ o ca v - u, o o ~ - ~ ~ ~ ~ o x ~ _ r~ 0- ~ 00 - - ~ oo o ~ ~ Zoo ~, r, ~} ~ _ Z Z ~ o ~ _ . . . . o o o o U. o =.= =- o~ C~ o ~ ~ ~ ~ ~ ~_ C~ . _ a.) o ~: ~ ~ e~ o 3 ~ o ~ _ o =.= O ~ CL c,) C) ~ . - ~: ._ l_ c~ cq ._ 1 I u~ c;S o C~ oo ~> :^ o ~ ~ ~ - . o o o ~ ~ - ) C ~ = - ~ X 3 ~ o O. C~ ~ o ~ ~ ~ D o U~ U~ >- ,Oq O O '_ O ~ C~ ~ oo ~i O V O E~ Z ~ ~ ~ D CD Ct o C~ C~ ~0 a' . ~ C~ a .e Ct o C~ o - S ~ Cq o ._ U, V, ._ C~ ;s C) - oo o C) ._ o V . . V o V)
173 and labor content in purchased components and materials; these data are presented in columns 2, 3, and 5.8 The calculation of U.S.-Japanese cost differences takes place in three steps. We first use the data in columns 2 and 3 to get an OEM hourly labor cost per vehicle of $1170, and then extrapolate using the cost shares (column 1) to arrive at a total manufactured cost and the cost of purchased components and materials (column 4~. Next, we multiply the cost per vehicle in column 4 by an esti- mate of the labor content of the three categories presented in column 5. The data imply, for example, that $1255 of the $1901 cost of components is labor cost. Finally, we calculate the U.S.- Japanese labor cost gap by multiplying the U.S. data in column 6 by 0.575; the adjustment factor is based on our previous estimate of the Japanese-to-U.S. unit labor cost ratio. 9 Thus, column 7 provides an estimate of the difference in the cost of producing a small vehicle in the United States and in Japan due to differences in unit labor costs, not only at the OEM level but also at the supplier level as well. The estimated cost gap is sizeable. When the effects of components and materials suppliers are added, the Japanese cost advantage is $1690. Although the calculations in Table A.1 are based on estimates of cost structure and labor content, reasonable adjustments of these assumptions would not reduce the order of magnitude of the Japanese cost advantage. To arrive at a landed (i.e., after shipment to the United States) cost differential, it is necessary to add general administrative and selling expenses as well as the costs of capital and transportation. Our estimates from annual reports and other sources suggest that these factors would reduce the Japanese advantage to $1436.~° The Company Perspective: Evidence from Annual Reports Additional insight into the differences in production costs in the United States and JaDan con he obt;~in.~rl 1-hr^~~rh :~n ::n~l~rcrie of · CALL ~ =~= V 1 ~ ~ · ~ · . _ Data contained In company annual reports. The use of annual reports shifts the focus of analysis to the costs incurred by the major manufacturers. In terms of labor costs the shift in focus generally means that no information will be available on labor embodied in components or materials. However, differences in costs associated with nonlabor inputs and with corporate-wide management and salaried personnel can be assessed. Further- more, the annual reports allow us to estimate labor productivity at the OEM level. This approach thus provides a useful check on industry estimates.
174 Any comparison of Japanese and U.S. companies must confront several analytical problems. Perhaps the most serious issue is the great difference in vertical integration and relationships with suppliers. At Toyota, for example, purchases account for almost 80 percent of the value of final sales. Because Toyota holds an equity interest in many of its suppliers, this figure is somewhat misleading. Comparable data for U.S. firms show much less reliance on suppliers. GM, for example, has a purchase-to-sales ratio of less than 50 percent. A second major problem is the different product mix of U.S. and Japanese firms. The data we shall use are those for 1979, when the products of the Big 3 were dominated by models in the medium-size ranges. The Japanese produced a much narrower range of vehicles, with heavy emphasis on the subcompact segment. Our approach to annual-resort analysis can be illustrated using data on Toyo Kogyo (Mazda) and Ford. Both companies provide sufficient information on automotive production and employment to permit calculation of labor hours per vehicle in U.S. and Japanese operations. Although Toyo Kogyo is about one-third the size of Toyota and used to be a relatively high-cost producer, it has experienced significant gains in productivity in recent years and now appears to have costs that are on par with those at Toyota and Nissan. Available evidence suggests that Ford is somewhat less efficient than GM, so that the Ford data may understate industry productivity.t ~ The basic estimates of employee costs per vehicle are presented in Table A.2. Two principal assumptions underly the calculations. Data on total domestic employment and total domestic employee costs were broken down into automotive and nonautomotive components based on the ratio of automotive to total sales. This effectively assumes that nonautomotive busi- nesses were as productive and as labor intensive as the automo- tive group. Since cars and trucks account for over 90 percent of sales at Ford and Toyo Kogyo, this assumption is not critical. The second assumption is that Ford employees worked an average of 1620 hours per year. For Toyo Kogyo the comparable number was assumed to be 1900 hours. These adjustments reflect differences in the effects of vacations, holidays, personal leave, and absenteeism. ~ 2 The evidence in Table A.2 reveals sizeable d ifferenc~ i n -a ~ > _ r_1 ~ . ~ ~ . ~ . . . . . _ . productivity and total employee cost per unit. Given our assump- tions, we estimate that the average Ford vehicle required 112.5 employee hours, while Toyo Kogyo produced an average vehicle in only 47. At an exchange rate of 218 yen to the dollar, we find that employee cost in the average Toyo Kogyo vehicle was less than $500; the comparable figure for Ford was $2464. The size- able cost gap reflects differences in product mix and vertical
175 TABLE A.2 Estimated Employee Costs Per Vehicle, 1979 Ford Toyo Kogyo (1) Domestic production of cars and trucks (millions) (2) Total domestic employment Automotive Nonautomotive (3) Total domestic employee hours Automotive (millions) (4) Total employee cost Automotive (millions) (5) Employee hours per vehicle (6) Employee cost per vehicle 3.163 219,599 19,876 $7794.50 112.5 $2464 0.9~83 24,318 2,490 46.20 $482.20 47.0 $491 NOTES: Line (1): Published production figures for Ford have been adjusted to eliminate 65,000 imported vehicles; the Toyo Kogyo data have been adjusted for production of knock-down assembly kits. Lines (2) - (4): Data on automotive employment and costs were obtained by assuming that the ratio of automotive employment to total employment was the same as the ratio of sales; the same assumption was made to obtain Ford employment costs. Line (3): Ford hours were determined by assuming that each employee actually worked 1620 hours per year. Toyo Kogyo hours assume that each employee actually worked 1900 hours. Line (4): Data include salaries, wages, and fringe benefits. For Toyo Kogyo, com- pensation data were derived by updating a 1976 figure using compensation growth rates at Toyota; an exchange rate of 218 yen per dollar (1979 average) was used to convert yen to dollars. integration as well as wage and productivity differentials. Toyo Kogyo concentrates heavily on the production of small cars, while Ford's product line covers a much wider range of sizes. Ford produces a larger fraction of the average vehicle in house. Infor- mation on value-added in the annual reports and discussions with industry sources suggests that the Toyo Kogyo results should be increased by 15-20 percent in order to adjust for differences in vertical integration. Using the higher estimates yields 56 hours per vehicle for Toyo Kogyo.~3 To correct for differences in mix we have estimated the cost to Ford of producing the Toyo Kogyo product mix. The calcula- tions are presented in Table A.3. The procedure uses data on manufacturing costs by vehicle size class developed for the Committee on Motor Vehicle Emissions of the National Research Council in 1974.~4 Estimates of the cost to Ford of producing the Toyo Kogyo mix were obtained by first computing a weighted average of the relative manufacturing cost indexes with Ford's 1979 production shares by size as weights. The ratio of the com- parable Toyo Kogyo weighted average (1.06) to the Ford weighted
176 TABLE A.3 Product Mix Adjustment (1 ) Ratio of car to total vehicle production (2) Production shares by size small medium large (3) Relative manufacturing cost by size (small= 1.00) small medium large (4) Weighted average relative manufacturing cost (small= 1.00) (5) Production of Toyo Kogyo mix at Ford level of integration (a) employee cost per vehicle (b) employee hours per vehicle Ford 0.645 o. 1 la o.68a 0.21 1.00 1.35 1.71 1.38 Toyo Kogyo 0.652 0.83 0.17 N/A N/A N/A 1.06 $1893b $589 87b 56 a Assumes that only Pinto and Bobcat models are small; Mustang and Capri sales were placed in the medium category. b Obtained by multiplying lines (5) and (6) in Table A.2 by (1.06 . 1.38). SOURCE: Committee on Motor Vehicle Emissions (1974); Ward's Auromotive Year Book; annual reports. average (1.38) was used to adjust both costs and productivity. It is an estimate of the effect of product mix on Ford's average _ cost and labor hours per vehicle. After these adjustments we estimate that Ford would require 87 employee hours to produce the average-size vehicle in the Toyo Kogyo product line, compared to 56 hours in the Japanese firm. Labor cost per vehicle is just over $1300 higher at Ford. These comparisons are based on the average-size vehicle at Toyo Kogyo. For a small vehicle (i.e., Pinto versus Mazda GLC) the Ford estimate is 82 hours per vehicle, while the comparable Toyo Kogyo figure is 53; the corresponding costs per vehicle are $1785 (Ford) and $556 (Toyo Kogyo). Even this adjustment may overstate costs and hours required to produce the Toyo Kogyo mix at Ford if the trucks and commercial vehicles produced by the two companies differ substantially. The analysis of annual report data suggests that the difference between Ford and Toyo Kogyo employee cost per small vehicle in 1979 was about $1200. Updating to 1980 would increase the absolute dollar difference by about 10 percent, a reflection of changes in wage rates and materials prices. An adjustment for
177 changes in exchange rates also would have only negligible effects. W e have used an exchange rate of 2 18 yen per dollar ( 1979) average; use of 200 yen per dollar (approximate rate at the end of 1980) would reduce the gap by about $50.~s The estimated cost differential reflects the compensation and productivity of all employees in the two firms. It does not, however, capture differences in unit labor costs in components or materials. Although the data were developed within a different f ramework, the evidence on labor content in components and materials used earlier is suggestive of the likely order of magni- tude. After adjusting for possible differences in productivity differentials at the supplier level, adding $700 to the employee cost differential seems justified. We conclude that differences in compensation and productivity lead to significantly lower unit employee costs in Japan; if Ford is indicative of average U.S. performance and if Toyo Kogyo is representative of the major Japanese auto producers (and comparisons with Nissan and Toyota suggest it is), then the cost advantage from labor and materials is likely to be about $1900 per vehicle. Analysis of other elements of total cost--e.g., selling and general administrative expenses-- provides evidence of a U.S. advantage of about $135 per vehicle. Likewise, freight costs and the 2.9 percent tariff add another $400 to Japanese costs. The net result is a landed-cost advantage to the Japanese producers of $1465. Evidence from the annual reports suggests a somewhat larger cost differential than we obtained using the industry/micro · ~ approach. The difference arises primarily from a much higher level of productivity in Toyo Kogyo than suggested by our updating of the 1974 estimates of Baranson. Those estimates applied to the entire motor vehicles and parts industry and were admittedly r ough and imprecise. It is not unreasonable that relative productivity at the OEM level would exceed levels achieved by suppliers. Without additional evidence it is difficult to judge the accuracy of the Toyo Kogyo data, yet similar calculations for the other major producers and discussions with industry experts suggest that the Toyo Kogyo analysis is representative. Indeed, care has been taken to make sure that our assumptions erred in the direction of underestimating the gaps in productivity and cost. Productivity at the Plant Level The final analytical approach involves a comparison of produc- tivity and cost on a plant-by-plant basis. The sources of these data are government surveys, plant visits by executives and engineers from the U.S. firms, and consultant reports." 7 Care
178 has been taken in these studies to compare similar processes and to correct for differences in product mix and degree of vertical integration. There is no claim made of complete coverage; only a few types of processes have been examined. It is felt, however, that several of the most critical elements in automobile produc- tion have been studied and that those studied are representative of the industry average. Table A.4 presents data on labor hours per vehicle in selected plants in the United States and Japan. It is evident that th e Jananese have a sizeahle nv`?rnil r~rncill~tivitv =Humnt~am anal th=+ ~ _ t~ } At_ ~ ~~ ~ ~~ ~] 1 Ill 1~ . ~ ~ ~ ~ . · · . . . ~ ~ ~ ~ . fine rental varies considerably in parts of the process. In or a,~ss~ons, foundry, and forge operations the differential ranges from 0 to 35 percent. The largest gap is in the stamping plants, where the Japanese advantage is close to 3 to 1. Stamping is one of the few processes where the Japanese appear to have a significant technology edge. The major press lines and transfer presses in Japan are equipped with U.S.-made automatic rolling bolsters and quick-die-change features. 8 Together with other automatic devices these features allow the Japanese to achieve output rates of 550 panels per hour versus 300-325 in the United States. The plant comparisons do not provide a complete set of cost figures, but the data can be used with previous information to obtain estimates of employee cost per vehicle. The comparisons imply that the ratio of Japanese to U.S. productivity at the OEM level is 1.9. Using an employee cost per hour ratio of 0.5, the productivity evidence implies a unit labor cost ratio of 0.263. If we apply this figure to our previous estimate of U.S. OEM employee costs per small vehicle of $1515, we end up with a TABLE A.4 U.S.-Japanese Difference in Productivity in Selected Plants: Hours per Vehicle Plant/Process Assem- Stamp- Trans- Country bly ing Engine mission Axle Foundry Forge Total United Statesa 38 10 7 8 5 5 1 74 Japanb 17 4 4 6 3 4 1 39 Difference 21 6 3 2 2 1 1 35 a SOURCE: J. E. Harbour, Comparison and Analysis of Manufacturing Productivity (final consultant report), Harbour and Associates, Dearborn Heights, Mich., 1980, p. 2. b SOURCE: Japanese Ministry of Labor, Statistical Survey of Labor Productivity, 1978, as cited in Abernathy et al. (1980, p. 41).
179 Japanese advantage of $1094. This assumes, of course, that the higher estimated productivity ratio is reflected only in lower Japanese hours, not in higher (and more expensive) U.S. hours. The procedure may thus understate the cost gap. The Impact of Capital The difference in labor productivity estimated here may be affected by differences in capital, which may in turn affect cost comparisons. Before examining evidence on this point it may be useful to clarify the issues with a simple diagram. Figure A. 1 (panel 1 ) presents two unit isoquants, one for the United States (Qus) and one for Japan (Qj). The vertical axis measures capital, and the horizontal measures labor. The isoquant depicts all the possible combinations of capital and labor that can be used to produce one unit of output. The way we have drawn the diagram implies that at a given capital-labor ratio the Japanese use less capital and labor to produce a unit of output than U.S. firms. This may be due to differences in management or techniques. We assume that the United States is at point A, with a capital-labor ratio given by the slope of the ray, OA. The J apanese are at point B. with a higher capital-labor ratio. At these points the Japanese have higher labor productivity and lower capital productivity. Assume for the moment that the unit isoquant for Japan in Figure A.1, panel 1, shifted upward to coin- cide with the unit isoquant for the United States, while Japan's capital-labor ratio remained unchanged. Japan would be at point C, while the United States is assumed to remain at point A. Compared with the United States, labor productivity in Japan is much higher, but capital productivity is lower. Depending on the prices of the inputs, total cost comparisons could go either way. Panel 2 uses the same type of diagram to illustrate a Japanese productivity advantage in both capital and labor. Here Q j is placed far enough below Qus that both less K and less L are required per unit, even though the K-L ratio is higher. To examine these issues we have developed estimates of capital-labor ratios and capital productivity using data from Ford and Toyo Kogyo for 1979. The estimates can therefore be compared to the productivity analysis for the two companies presented earlier. The results of our calculations are presented in Table A.5. Any attempt to compare capital stocks in two countries must confront problems of inflation and differences in prices and currency values. Both problems are addressed in Table A.5, but we also present unadjusted values for comparison. The annual r - - - ,,
180 - <~ K fit: y - ~ KUs <: Kj Panel 1 - / J Lj \ LABOR (L) >< ~ , _ _ I' _ / Imp< 1 . 1 Lus Panel 2 J US Lj LABOR (L) 'US 'US FIGURE A. 1 Capital-labor ratios: United States and Japan.
181 TABLE A.5 Capital-Labor Ratios and Capital Productivity, 1979 Toyo Kogyo General Specific Exchange Exchange Category Ford Ratea Rateb Plant and Equipment (billions of dollars) Book value Adjusted for inflations Inventory (materials and work in proces~billions of dollars) Capital-Labor Ratios (dollars) $15.330 $1.539 $1.269 26.975 2.509 2.068 $2.939 $0.077 $0.077 Capital per employees Book value $40,063 $ 63,113 $52,003 Adjusted 65,599 100,888 83,128 Capital per labor hours workede $24.73 $33.22 $27.37 Adjusted 40.49 53.10 43.75 Capital Productivity (dollars) Capital per vehiclef Book value Adjusted $3,048 $1,639 $1,351 5,052 2,623 2,161 a Average 1979 exchange rate was 220 yen per dollar. b Capital information exchange rate (structures and equipment) calculated based on data in Irving B. Kravis et al., International Comparisons of Real Product and Purchasing Power (Baltimore, Md.: The Johns Hopkins Press, 1978); 267 yen per dollar in 1979applies only to gross plant and equipment. CTo correct plant and equipment for inflation, book values were restated in 1979 prices based on age of capital and variation in equipment and structures deflator from national in- come accounts; only estimated capital used in vehicle production is included (calculated based on fraction of total sales in automotive). Capital includes both gross plant and equipment and inventories; no inflation adjustment is made for inventories. e Annual hours per employee are 1620 for Ford and 1900 for Toyo Kogyo. f Vehicles include cars and trucks; excludes tractors for Ford. SOURCE: Annual reports, 1979. Deflators were taken from "Monthly Finance Review," Japan Ministry of Finance (Dec. 1980); "Quarterly National Accounts Bulletin," OECD; "Historical Statistics" OECD; Economic Report of the President, 1981, Council of Eco- nomic Advisors. Purchasing power parity exchange rates were taken from Irving B. Kravis et al., International Comparisons of Real Product and Purchasing Power (Baltimore, Md.: The John Hopkins Press, 1978).
182 reports for 1979 provide data on the gross book value of plants (including land) and equipment and the value of work in-process inventories. The yen values for Toyo Kogyo have been translated into dollar values in two ways. The first uses the general _= ~ r _~^ r I I ex<:nange rate of ~ flu yen per Molar, while the second uses a specific or purchasing power parity exchange rate of 267 yen per dollar. This latter measure is derived by comparing the yen and dollar prices of comparable (in terms of quality, performance, function, etc.) equipment and structures at a point in time. We have used the estimates prepared by Kravis et al. for capital formation (structure and equipment) in 1973, updated by differ- ences in the rates of inflation in the equipment and structures deflators from the United States and Japan national income accounts. The results for book value show Ford with 15.3 billion dollars, and Toyo Kogyo with 1.54 billion dollars or 1.27 billion dollars depending on the exchange rate used. Inflation affects comparisons because of the accounting convention of reporting assets at original cost. We have restated the book values in 1979 dollars using rough estimates of the age of the capital stock in the two companies and the business fixed investment deflators from the national accounts. The formula can ne written as follows: GBVadj = GBV rP(1979) L P(1979 - Age) where P is the investment deflator, the number in parenthesis is the year in which the price index is measured, and age is calcu- lated as the ratio of accumulated depreciation to annual depreciation. Since the ages of the capital stock at Ford (8.1 years) and Toyo Kogyo (9.4 years) are comparable, and since the rates of inflation have not been greatly different, the inflation adjustment has only a small effect on the relative amounts of plant and equipment. The data on age of capital suggest that U.S. and Japanese auto firms are using equipment of comparable vintage. it does appear, however, that the Toyo Kogyo data may overstate the average age of the capital stock for all Japanese automobile manufacturers. Data on Nissan, for example, show an average age of capital of 7.1 years in 1979. Likewise, the Ford data may understate the average age of the capital stock for U.S. automobile manufacturers. Data on G M, for example, show an average age of capital of 11.4 years in 1979. These data imply that the capital stock in the U.S. auto industry is, on average, 3-4 years older than the capital stock in Japan. For comparisons made in 1979 this implies that the average piece of equipment in Japan was purchased in 1972, while the average piece of equipment in the U.S. firms was purchased in 1968-1969.
183 If there were significant technological breakthroughs between 1968 and 1972, the difference in age of capital between the U.S. and Japanese auto industries could imply important differences in the level of technology. Research by Abernathy on the develop- ment of the production process in the auto industry suggests that technology in the late 1960s and early 1970s was not character- ized by major changes.~9 While a difference of 3-4 years in the age of the capital stock is likely to be of some importance in productivity comparisons, the implication of existing evidence on the age of capital and the course of technical change in the auto industry is that the level of technology embodied in plant and equipment in the U.S. auto industry is not greatly different from the level found in the auto industry in Japan. However, more detailed research on the nature and timing of technological change in the auto industry is needed before strong conclusions can be drawn. After adding material and work in-process inventory (which is not inflation adjusted) to gross plant and equipment, line III presents measures of capital-labor ratios. We f irst calculate capital per employee and then take differences in annual hours worked into account. The data support the conclusion that Toyo Kogyo operates with higher capital-labor ratios, irrespective of the measure of labor input used. The estimates of the Toyo Kogyo edge range from 58 percent for the unadjusted book values per employee to 8 percent after correcting for inflation, purchasing power parity, and differences in annual hours worked. These conclusions are unchanged if data for 1978 or 1980 are used instead, implying that adjustment for utilization differences does not affect the results. The important point to note is that the differences in capital- labor ratios, even without any adjustments, are not sufficient to explain the productivity gap observed earlier. As far as differ- ences in OEM hours per vehicle are concerned, the effect of capital depends on capital's share in the cost of value-added. Even if this were as high as 0.5 (for the economy as a whole it is more like 0.25-0.30), capital differences might explain a labor productivity difference of 30 percent, rather than the difference of 55 percent we actually observe. If we use a capital share of 0.3, the potential impact of differences in capital-labor ratios between the United States and Japan on labor productivity ranges from 2.4 to 17.3 percent, depending on the specific definition of capital and labor used. The evidence in line III suggests that differences in capital- labor ratios cannot fully explain the productivity gap. It appears that more is involved than substitution of capital for labor. Line IV underscores this point by comparing capital per vehicle. As they stand, the estimates suggest that panel 2 of Figure A.1 is a
184 TABLE A.6 Summary of U.S.-Japanese Cost Differences Approaches Cost Category Industry/Macro Industry/Micro Company Plant (per vehicle) (1) (2) (3) (4) OEM employee cost N/Aa $871 $1219 $1115 Materials and components N/Aa $819 $700b $700b Other costs N/Aa $146C $146C $146C Freight and duties N/Aa ($400)~ ($400) ($400) Landed-cost difference $748 $ 1436 $ 1665 $ 1561 a These data are not included in the analysis; sources used only provided a total cost differ- ence. b This estimate starts with the $819 difference in column 2; it is assumed, however, that productivity in the supplier sector is lower than the OEM level and that the wage ratio re- ma~ns unchanged. As a first approximation the material advantage has been reduced to $700. c Other costs include capital charges, general selling and administrative (GS&A) expenses, warranty costs, and costs of in-country transportation. The capital costs are taken from Table A.5 as described in the text; GS&A expenses have been taken from annual reports. The breakout is as follows: capital = $68; GS&A = ($132). The Japanese have an advantage in warranty of $90; their inland freight is $120 less expensive. These data are taken from a re- port by Harbour Associates as reported in Abernathy et al. (1980, p. 60). Numbers in parentheses indicate the U.S. advantage. more accurate reflection of the situation. The calculations imply that Toyo Kogyo uses less labor and less capital per vehicle. While these data have not been adjusted for vertical integration or for differences in product mix, doing so would not affect the conclusion. If we use the same adjustment for product mix that we used in the labor productivity analysis (see Table A.3), we would reduce the Ford book value per vehicle from $3048 to $2340 (the adjustment factor is 0.768~. If we also use the same adjust- ment for vertical integration that we used in the labor produc- tivity analysis, we would increase Toyo Kogyo's book value per vehicle by 15 percent, from $1639 to $1885. The product mix adjustment is likely to understate capital per vehicle at Ford, because capital input does not increase with the size of the vehicle as rapidly as labor and materials.20 Even so, these adjustments show that the Japanese firm uses less capital per vehicle than its U.S. competitor.
185 The evidence in this table reinforces the observation of the panel experts that differences in technology and automation, while factors, were not the most important determinants of differences in labor productivity. Moreover, the table adds further evidence that the overall cost and productivity gap lies in the more effective and efficient use of relatively comparable resources. If we use capital per vehicle adjusted for inflation and the specific exchange rate and assume a capital charge (depreciation plus interest) of 15 percent, the numbers in Table A.5 imply a Japanese advantage of $433 per vehicle in capital costs t($5052 - $2162) x 0. 153. If the Ford capital per vehicle is reduced to reflect differences in product mix as described above, and if the Toyo Kogyo capital per vehicle is increased by 15 percent to reflect differences in vertical integration, we end up with a Japanese advantage of $209 ~$3880 - $2485) x 0.153. Finally, if we use book value per vehicle and the general exchange rate, and apply the product mix adjustment and the adjustment for vertical integration, the result is a Japanese advantage of $68 t(52340- $1885) x 0.153. This is the value we have used in the overall cost comparisons. Because the product-mix adjustment may under- state capital per vehicle at Ford, and because we have used the general exchange rate and book values, the $68 estimate may understate the Japanese advantage. Summary Table A.6 summarizes the cost differences obtained under the four approaches and adds estimates of other costs, indirect freight charges, and corporate overhead. These calculations are neces- sarily rough. And we have not maintained strict independence, since a given method may use information obtained under a different perspective. As noted throughout this appendix, where assumptions were necessary we have tried to err on the side of understating the cost advantage of the Japanese. Nevertheless, the results point to a significant differential ranging from $1000 to more than $1400. NOTES 1. These estimates are based on unpublished work done at the Research Department of the United Auto Workers. 2. Baranson's estimates are presented in Toder (1978), p. 151. 3. British Central Policy Review Staff (1974~.
186 4. These estimates are based on data from the Japanes e Ministry of Labor and the U.S. Bureau of Labor Statistics. 5. This figure is based on unpublished data from the U.S. Bureau of Labor Statistics (BLS), as well as published company information (annual reports, etc.~. Fringe benefits differ significantly between the U.S. and Japanese auto industries, and we have attempted to reflect these differences. Given the importance of "internal" friezes in JaDan. there maV he errors: in a, . . ~ . one estimates. Discussions with knowledgeable participants in both the United States and Japan suggest that an>, errors are minor. For 1980 the BLS estimated the U.S. and Japanese compensation rates in the motor vehicles and equipment industry as follows: Average Hourly Earnings United States $9.81 $14.71 Japan $5.96 $ 6.98 Hourly Compensation Source: Unpublished data, U.S. Bureau of Labor Statistics, October 1981. 6. Note that the calculations are based on compensation and not the costs to the employer of a unit of labor. The ratio ignores costs to the employer for an hour of work that does not show up in the employees' direct compensation--taxes, absenteeism, and so on. Since these terms tend to be larger in the United States. the calculations may understate the Japanese advantage. It should be noted that the panel did not deal with the question of relative rates of compensation for top executives in the U.S. and Japanese auto industries. Although compensation data have been used for all salaried workers in the cost comparisons, no attempt was made to break out executive compensation (i.e., vice-president and above). The view that top executives in Japan have lower rates of pay than their U.S. counterparts has been expressed in the litera- ture, but the panel had no basis for judging that claim. The absence of disclosure requirements and the complexity of top- executive pay make comparisons difficult. Furthermore, U.S.- Japan differences in top executive compensation are likely to have only minor effects on cost comparisons. For example, in a good year, such as 1978, the top five executives at Ford were paid an average of 5638,000 in compensation. Assume for the moment that the top 25 executives at Ford were all paid this amount, while their counterparts at Toyo Kogyo were paid nothing. Such a difference in executive compensation would add only $6 to the per vehicle cost difference between Ford and Toyo Kogyo. (This assumes that all executive compensation applies only to U.S. passenger car production, which was about 2.6 million in 1978.)
187 7. Committee on Motor Vehicle Emissions (1974~. Note that we have assumed an average level of options. Industry sources include staff reports of (J.S. companies, memoranda from members of the panel, and informal discussions. 8. The nominal cost per hour worked is the cost to th e employer and includes base rates, fringes, and other payroll costs. We assume that material suppliers (steel, plastics, etc.) have the same ratio of unit labor costs as participants in the industry. 9. Let C(US) and C(~) be unit labor costs in the United States and Japan, respectively. We estimate C(~/C(US) = 0.425. We want to know C(US) - CtJ). Column 6 gives us C(US). Thus, C(US) - C(UJ) = t1 - CtJ)/C(US)] x column 6; this result is column 7. 10. See notes to Table A.6 for a breakdown. These estimates pertain to 1980; updating to 1981 would raise them somewhat. 11. This is documented in Abernathy et al. (1980~. 12. The data for annual hours worked for Ford assume 35 days of contractual vacation, paid personal holidays, and so on and an adjustment for absenteeism of 5 percent of normal annual hours (assumed to be 2000~. The Toyo Kogyo hours are based on 2037 hours average in Japanese manufacturing, plus adjustments based on holidays and vacations. 13. The vertical integration ad justment reflects industry judgement rather than analysis of purchases-to-sales ratios; on the latter basis the two companies are comparably integrated. 14. The procedure probably understates the number of hours per vehicle at Ford, because the adjustment is based on total manufacturing costs, not the cost of labor alone. 15. If applied to total passenger car production, the large cost difference reported here would seem to imply much higher profits than Japanese firms typically report. Since accounting practices differ between the two countries, it is difficult to interpret such apparent discrepancies. One possible explanation is the difference between profits for Japanese auto companies in their home market and profits in the U.S. market. Discussions with Japanese executives suggest that the U.S. market is much more profitable than the domestic market in Japan. The overall level of Japanese profits, therefore, may be an average of very profitable and only marginally profitable markets around the world. 16. If the industry data on productivity are accurate (Japan-to-U.S. ratio of about 1.2), and we use the Toyo Kogyo-Ford ratio at the OEM level (about 1.5), then the implied level of productivity for suppliers and the assumed wage ratio justifies a differential of about $650-$700 per vehicle. 17. Plant-level data on Japan are available from the Japan Ministry of Labor in its annual Labor Productivity Statistical Survey. U.S. data have been obtained from industry sources. For additional data, see Abernathy et al. (1980~. 18. Abernathy et al. (19sn)
~ go 19. See Abernathy (1978), pp. 86-113, for a discussion of technology in engine plants and pp. 114-J46 for a discussion of assembly plant technology. 20. For evidence on this point, see Abernathy (1978~, pp. 21 and 193-194' where the unit body construction method used on small cars is described. It appears that small-car production may actually be more capital intensive (more capital per vehicle) because of many fewer parts and unit construction.
