The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
( e c o n o m i c REVIEW I ---------------— -------------FEDERAL RESERVE BANK | of CLEVELAND .— _ _ _ _ _ _ _ _ _ _ — -------------- SUMMER ! ,i I Economic Review Summer 1984 The Implementation of 2 Industrial Policy ........ Proposals for a coordinated industrial policy, designed to enhance the productivity and competitive position of the U.S. economy, have attracted many supporters. Industrial policy will require the creation of an agency to plan and execute governmental directives. Using the American experience with national economic planning during two World Wars and the New Deal, economist Daniel A. Littman discusses the ability of government to imple ment such a policy. The author identifies serious problems that may frustrate govern ment attempts to implement an industrial policy for the 1980s. Voluntary Export Restraints: The Cost of Building W a lls................. 17 At the urging of the U.S. government, the Jap anese restricted exports of new cars to the United States in 1981. Designed to protect jobs in the U.S. auto industry, these voluntary export restraints (VERs) have done so at the expense of American consumers. Economists Michael F. Bryan and Owen F. Humpage derived estimates of the restraint program’s cost and its contribution to U.S. auto-related employment by developing a model of the mar ket for Japanese cars in the United States. The model incorporates dealers’ inventory positions and allows for variable dealers’ mark ups over options-adjusted wholesale costs. Economic Review is published quarterly by the Research Departm ent of the Federal Reserve Bank of Cleveland, PO. Box 6387, Cleveland, OH 44101. Telephone: 216/579-2000. Editor: Pat Wren. Assistant editor: Meredith Holmes. Design: Jamie Feldman. Typesetting: Lucy Balazek. Opinions stated in Economic Review are those of the authors and not necessarily those of the Fed eral Reserve Bank of Cleveland or of the Board of Governors of the Federal Reserve System. M aterial may be reprinted provided that the source is credited. Please send copies of reprinted m ateri als to the editor. ISSN 0013-0281 Daniel A. Littman is an economist at the Federal Re serve Bank of Cleve land. The author would like to thank Robert Cuff, Paul Koistinen, Barry Mitnick, Bill Pierce, Stephen Skowronek, Ed Stevens, and Gary Wyckofffor their insightful com ments on drafts of this article. The Implementation of Industrial Policy by Daniel A. Littman Federal Reserve Bank of Cleveland A growing coalition of U.S. political, labor, and business leaders is lobbying for a coordi nated interventionist national industrial policy. This coalition advocates measures designed to enhance the competitiveness and productivity of the U.S. economy. The term industrial policy means various things to var ious people, a fact that manifests itself in the wide variety of policy proposals under consideration. Industrial policy enthusiasts agree only that the U.S. economy is expe riencing grave structural problems and that the federal government has both the obliga tion and the capacity to implement policies to alleviate these problems. The industrial policy proposals range from measures intended to enhance the long-run efficiency of competitive markets, to steps to allow the federal government to undertake decisions normally handled by the private market. Some proposals include modifi cations of the tax code to promote research and development activities, and relaxation of antitrust guidelines to permit consolidation of production facilities in troubled industries. More common are proposals with such ambi tious elements as establishing a governmentfinanced Development Bank to extend pref erential credit to so-called sunrise and sunset industries; forming a nonpartisan council to supervise the bank’s activities and coordinate the industrial policy initiatives of govern ment agencies; and restricting interstate bus iness mobility and corporate diversification. The existing critical literature on indus trial policy focuses on diagnosing the struc tural problems of the U.S. economy and on the success or failure of domestic sectorspecific and comprehensive foreign industrial policies (see box 1). However, even if a con sensus were to develop regarding the sources of and remedies for the U.S. economy’s struc tural ailments, industrial policy still faces an uncertain transition from law to delivery— 1. See Pressman and Wildavsky (1973), Mazmanian and Sabatier (1983), and Peirce (1981). a process known as implementation. In this article, our concern is the government’s abil ity to implement a complex and coherent array of government actions under the umbrella of industrial policy. Any activity engaged in by a democratic government requires policy choices by elected representatives and their appointees, as well as policy execution by civil servants. The activities of those charged with implement ing policy—detailed design, delivery, perfor mance monitoring, and enforcement—have been analyzed extensively, and they have been found to be afflicted by predictable lim itations and imperfections.1 The ability to implement policies legislated by a govern ment frequently is neglected in policymaking and analysis. Seeming to believe that imple mentation is a relatively easy task, legislators invariably are upset when policy goes awry, or when policy results are contrary to legisla tive intentions. When the objectives of policy are not realized, the possible causes are Box 1 Critical Views on Industrial Policy The critical literature on industrial policy has focused on two sets of issues. First, analysts have questioned the extent to which the U.S. economy is experiencing grave structural problems. Advocates maintain that the structural difficulties of the U. S. economy are reflected in the following phenomena: (1) a progressive deindustrialization that could harm m anufacturing industries and union workers in the short run and could threaten national economic growth in the long run; (2) a secular deterioration in productivity, capital invest ment, and international competitiveness; (3) a national capital market that discriminates against infant industries, research and development ventures, and investm ents in public infrastructure. Since these assertions have been analyzed in recent literature, it is not our purpose here to provide an exten sive review. Critics cast doubt on assertions that the U.S. economy is burdened with unparalleled difficulties of a structural nature. The literature also suggests that the government would, even under ideal conditions, lack the inform ation necessary to make responsible and potentially effective policy decisions? Second, researchers have examined comprehensive foreign and domestic sector-specific industrial policies Economic Review • Summer 1984 many. The fault could lie in a misdiagnosis of the problem, or problems, to be solved. Legislative intent might have been defective: the objectives of policy may have been unat tainable, or the policy targets poorly selected. Another explanation may be that implemen tation was faulty: the mandated activities were improperly executed or experienced damaging delays. Policy implementation is complex. It begins with the passage of legislation; works through establishing an administrative structure to initiate policy and select objectives, targets, and instruments; and ends with delivery. The process involves an intricate web of rela tionships among policy advisors, legislators, executive branch officials, and superiors and subordinates within the agencies charged with execution. It must be recognized that gov ernments and their component bureaucratic institutions are not monolithic (Downs 1967), a fact of considerable importance in the translation of policy directives into policy used as models by advocates. They maintain that indus trial policy was a crucial factor in the rapid postw ar grow th in Japan, West Germany, and France. Critics point out that such research has failed to dem onstrate a causal relationship running from industrial policy to economic growth. Using standard and quantifiable eco nomic, cultural, and demographic measures, advocates have not been able to explain all of the postw ar eco nomic grow th of the three countries. The often large and unexplained residual is then attributed to the effec tiveness of industrial policy.b Regarding sector-specific industrial policies in the United States, Nelson (1982) argues that failures have been at least as common as successes, and that success has occurred under special circum stances, suggesting that policy replicability may be limited? For instance, government support of research and development in American agriculture (the quintessential sector-specific policy) is a poor model of centralized planning for the m anufacturing sector. a. See Daniels and Kieschnick (1978) and Industrial Change and Public Policy (1983). b. See Maunder (1979), Trezise (1983), Burton (1983), and Pavitt and Walker (1976). c. See Nelson (1982) and Wescott (1983). 2. For further dis cussion of the prin cipal/agent model and its applications, see Mitnick (1974), Mitnick and Backoff (1983), and Banfield (1975). 3. The potential implementation problems of modern industrial policy might be identified via three avenues, only one of which— historical evidence— is pursued in this article. The imple mentation difficul ties encountered by policymakers in Japan, France, and West Germany could be relevant, despite the substantial dif ferences in culture, political systems, and economic struc ture between these nations and the United States. Alter natively, the imple mentation experience of domestic sectorspecific industrial policies could be examined for appli cable problems. outcomes. Within each bureau, for instance, high echelon officials are charged with the design and execution of policy directives. Because officials consider numerous policies, each single policy is defined in general terms. Deputies are left to work out the details. These deputies, in turn, typically delegate much of the detail work to individuals below them in the organizational structure. Orders from the top are expanded and made more specific as they move downward. Since each participant has some discretion in selecting alternatives, the policies of an organization (or a government) are defined at all levels, not simply at the top. The theory of agency has been employed by researchers from a variety of social and behavioral science disciplines to examine the kind of complex problems that surface in the implementation process. An agency relation exists when one party, the agent, is acting for another party, the principal. The agent gen erally is construed to be acting for the benefit of the principal. Since the principal is sep arated from the activities conducted by the agent on his or her behalf, he or she must establish systems of monitoring and control to assure that the agent behaves as desired.2 The implementation process consists of many layers of responsibility and delegation. In this article we examine industrial policy based on three layers of delegation, from objective-setting to the delivery of policy. The topmost officials, including the president, cabinet members, and congressional commit tees, are the principals for whom industrial policy is implemented. Although they often possess contradictory policy preferences, the principals must provide the organizing direction and general objectives of policy to agents employed to design and execute policy. The officials and staff of the industrial policy bureau, a second group, are the primary Federal Reserve Bank of Cleveland responsible for developing a detailed policy design and directing the instruments of policy. For industrial policy to be poten tially effective, it required that another layer of secondary agents—the individual firms and plants in each regulated industry—actively buy into and participate in the implemen tation process. In the implementation of public policy, prin cipals must concern themselves with four agent-control strategies, to assure that the policies implemented conform to initial direc tives. First, principals must determine the degree to which the policy preferences of their selected agents diverge from their own. The combination of differing preferences, a normal amount of responsibility delega tion, and agent discretion concerning policy alternatives can result in policies that bear little resemblance to the principals’ original directives. Second, principals must transmit detailed, intelligible, and feasible policy speci fications to their agents. Agents may then be expected to discriminate between actions that are desirable and undesirable in the prin cipals’ eyes. Third, principals must be able to monitor the behavior and performance of agent actions on their behalf, by establishing information feedback mechanisms and by identifying performance measures that pos sess a predictable relationship to the desired final policy outcomes. Finally, principals need to formulate a system of rewards and penal ties to help assure that agents are implement ing the proper policy with a minimum of delay and deviation. The existing literature on industrial pol icy overlooks the implementation process. Potential barriers to industrial policy cannot be analyzed directly, since examples of coordi nated and interventionist policy envisioned by advocates are not in place to provide a basis for study. The history of U.S. economic interventions, however, provides a useful laboratory for examining industrial policy.3 agents In three episodes of national crisis during the twentieth century, the federal government established powerful executive agencies to coordinate and direct a broad range of pri vate economic activities. Today’s proposals envision a similar agency. In World War I, President Woodrow Wilson formed the Coun cil of National Defense (CND) and the War Industries Board (WIB) to mobilize the pri vate economy for war. During his celebrated first 100 days in office, President Franklin Roosevelt formed the National Recovery Administration (NRA) to plan and stimulate the private economy. Finally, early in our involvement in World War II, President Roosevelt established the War Production Board (WPB) to plan and supervise our industrial and transportation sectors. Each of these agencies closely resembled, in struc ture and responsibilities, the proposed delivery agency for contemporary industrial policy. None was designed to serve exclusively parti War Industries Board Sudden involvement in World War I forced the first U.S. attem pt at economic and social control on a com prehensive national scale. For almost two years, the national governm ent sought to command the major industrial and manpower resources of the United States; recruited, trained, and armed an expedition ary force; and served the financial, ordnance, and eco nomic needs of our European allies. The governm ent’s wartime economic powers required some central coordi nation, although public officials had no familiarity with large-scale economic planning. Established in 1917, the Council of National Defense (CND) and the War Indus tries Board (WIB) were charged w ith the coordination of industries, resources, and transportation facilities for the national security and welfare. The WIB replaced the m arketplace where large industry was concerned, employing persuasion, threats, and calls to patriotic duty to gain “acceptable” prices, adjustm ent of com peting claims for scarce resources, and priorities in m ilitary and civilian contracting. Among its specific duties were advising the president, executive agencies, and Congress on appropriate and necessary courses of action to mobilize the private economy; coordinating transportation, communication, and production facili ties; acquiring strategic raw materials; establishing and enforcing priorities in production and delivery of w ar Economic Review • Summer 1984 san interests, and senior staff included rep resentatives from government, business, labor, and the public. All three were assigned broad responsibilities to direct and coordinate the related activities of other executive agen cies. Each was associated with a government agency whose responsibilities included lending to the private sector. Finally, all three were conceived as clearinghouses for information, research, and planning of public and privatesector activities. I. Agents and Institutional Capacity The enabling legislation for the WIB, NRA, and WPB gave unprecedented and unfamiliar powers to the federal government. In the world of public policy, there is, however, no assurance that implementation will be effec tive, or that governments will only choose policies that feasibly can be executed. Institu- materials; fixing prices of military and civilian goods ranging from bullets and machine guns to coal and grain; directing the conversion and/or construction of m anu facturing and transportation facilities for defense pro duction; and coordinating the purchases of Allied gov ernm ents in the United States. The WIB worked closely w ith other agencies established in the national em er gency, including the Emergency Fleet Corporation, the United States Railroad Adm inistration, the U.S. Food and Fuel administrations, the War Finance Corporation, and the National War Labor Board. The WIB was dis solved in December 1918, just one month after the w ar ended. While the WIB established a semblance of order in economic mobilization by the sum m er of 1918, the agency never secured unified control over that mobili zation. It lacked clear authority to set civilian and mili tary prices and to coordinate transportation, duties that were covered, in part, by other agencies. Through out the war, its officials rem ained unsure of their legal authority to coerce businesses and to enforce contracts w ith m ilitary agencies. The WIB’s rapid dissolution prevented wide public aw areness of the serious w eak nesses and inconsistencies of government economic controls during the war. For greater detail on the WIB and its sister agencies, see Cuff 1973, Garfield 1921, and Willoughby 1934. 4. See Skowronek (1982) and Fine (1956). For com parisons with the European political and institutional tradition, see Hartz (1964) and Batchelder and Freudenberger (1983). tional capacity, defined here as a combination of the government’s organizational manage ment skills and technical knowledge, is an important influence on the speed of the implementation process and the potential for successful outcomes, particularly with respect to new policy directives. Finegold and Skocpol (1982, p. 260) observe, “Governments that have, or can quickly assemble, their own knowledgeable administrative organizations are better able to carry through interven tionist policies than are governments that must rely on extragovernmental experts and organizations.” In addition, new agencies are usually influenced disproportionately by the regulated groups that lobbied for their creation. The interventions authorized by the National Recovery Administration Created in 1933 under the National Industrial Recovery Act, the NRA was modeled on the WIB of World War I and included key personnel of the WIB among its staff. The NRA’s prim ary feature was the organization and adm inistration of industrial self-government, suspend ing antitru st enforcement and allowing governmentsponsored cartelization of industry. Elaborate codes of business behavior for employment, investment, output, wages, and prices were to be established by mutual agreement. Eventually, over 500 individual industry codes were developed, and signatories to these agree m ents were allowed to display the NRA blue eagle insig nia. While the NRA lasted for two years and vanished w ith hardly an institutional trace, it was the center piece of government recovery policy in the early New Deal. The codes were intended to encourage cooperation that would eliminate overproduction and the resulting downward pressure on prices and wages—that is, to stem deflation. Conservatives believed that the end of destructive competition, as they viewed it, would bring optimism and a renewal of private investm ent. More am bitious proponents reasoned that if codes could be drafted to keep prices down but deliberately increase wages, industry, in effect, would be forced to come up with the money to revive U.S. purchasing power. Having failed for many reasons, some of its own making, the NRA was repudiated by many of its early supporters by the tim e the Supreme Court declared it unconstitu tional in 1935. For greater detail on the NRA, see Finegold and Skocpol 1982, Bellush 1975, Hawley 1969, and Brand 1983. 6 Federal Reserve Bank of Cleveland industrial policy legislation of 1917, 1933, and 1939 presented the federal government with serious technical and organizational capacity problems. No ready-made administra tive structure existed within the federal gov ernment to design and implement such pol icies. Compared with our allies in Western Europe, the federal government was handi capped by a historical tradition that empha sized decentralized decisionmaking and “anti-statist” rhetoric.4 Elected representa tives, civil servants, and political appointees did not have the vast technical and bureau cratic knowledge necessary to design and implement industrial policy. By World War I, the evolution of large private companies and industry trade associations had given some of these important skills to the private sector (Chandler and Galambos 1970). The Wilson and Franklin Roosevelt administrations desired rapid implementation of industrial policy, but could not easily assemble from the government’s own resources either the nec essary administrative apparatus or the tech nically skilled personnel needed to direct and staff the agencies. In what should be consid ered classic cases of regulation by the regu lated, both presidents turned to the private sector for their industrial policy agents. The CND and the WIB were essentially exercises in improvisation. The proposals to establish the two agencies were put forward by a coalition of major defense contractors and other private businessmen and did not originate with the president or Congress. Once the agencies were operational, initia tives for administrative change and new powers came, once again, from the business community. President Wilson used his con siderable emergency powers to work out interim agency arrangements with private- sector interests. The WIB was directed and staffed primarily by businessmen, financiers, and trade association lobbyists from the pri vate sector, although labor and government had token representation. The WIB was disbanded immediately after World War I, with its staff returning to privatesector employment. The Harding and Coolidge administrations dismantled most of the formal institutional legacy of the World War I era (with prominent exception of informal planning activities conducted by the newly War Production Board The mobilization of the U.S. economy in World War II was an achievement w ithout precedent in magnitude, complexity, or duration. The defense mobilization and production requirem ents greatly exceeded those of World War I, and w rought sweeping changes in the operations of government and in the structure of the U.S. economy. The federal government established a num ber of agencies w ith overseeing authority over the huge effort, beginning w ith the War Resources Board (1939), the National Defense Advisory Commis sion (1940-41), the Office of Production M anagement (1941-42), and the Supply Priorities and Allocations Board (1941-42). The evolution toward a centralized economic planning agency culminated in the War Pro duction Board (1942-45). The WPB had considerably greater powers to mobilize the national economy than either of its predecessors—the WIB or the NRA. It was given authority to divide resources between civilian and military claim ants, establish production and deliv ery priorities w ithin the two groups, coordinate and oversee the building of production facilities and the con version of existing facilities, and coordinate manpower needs. The WPB was dissolved shortly after the cessa tion of hostilities in the Pacific theatre in August 1945. The WPB’s responsibilities were often ambiguous, and they frequently overlapped with the jurisdictions of other emergency agencies, w ith predictable confusion resulting. The demands of World War II on the U. S. econ omy and on the governm ent’s planning agencies were so great that even an abundance of internal and exter nal experts could not fashion a coherent set of govern ment institutions or policies. The economic planning agencies of World War II were rarely effective as long term strategic planners, but relatively successful at crisis m anagem ent. For greater detail on the WPB, see C ivilian P roduction A d m in istratio n 1947, N ovick, Anshen, and Truppner 1949, Koistinen 1980, and Rockoff 1984. Economic Review • Summer 1984 formed U.S. Department of Commerce in the 1920s). The federal government thus faced the Great Depression with little more inter nal administrative and technical capacity than it possessed in 1917. President Franklin Roosevelt and his advisors employed the WIB as a model for the NRA, asking several WIB veterans to form the staff nucleus of the agency. Although the staff included represen tatives of government and organized labor, the NRA came to reflect the interests of large U.S. private corporations and their execu tives. The deputy administrators were drawn almost entirely from the ranks of business, and were often assigned to direct code negoti ations with their own industries. Likewise, the code authorities formed to monitor com pliance were staffed by business executives and by trade association personnel (the NRA promoted the formation of trade groups in industries with no pre-existing association). The U.S. government entered World War II with a large organizational structure, cour tesy of the New Deal. The government, how ever, did not have the technical expertise necessary to prepare the private economy for war. The NRA was disbanded in 1935, and major segments of the business community were alienated from the Roosevelt program. The WPB staff was recruited from the pri vate sector, and often conducted procurement and price-fixing negotiations with their own industries. The federal government established princi pal/agent relationships to implement indus trial policy in 1917, 1933, and 1939, largely as a result of technical and organizational defi ciencies. Such arrangements are (and were) not unusual in public policy: the relation between the president and government bu reaus staffed by professional civil servants 5. Cuff (1973), Koistinen (1980), Solo (1959), and Christman (1973) show that the WIB and WPB had fre quent policy dis putes with the pres ident, his cabinet, and congressional com mittees responsible for agency oversight. Serious conflicts took place between the two agencies and other bureaus of government, espe cially the military procurement depart ments. Himmelberg (1968) discusses the attempt, by WIB supporters, to estab lish a peacetime price-fixing agency (the Industrial Board of the Department of Commerce) in 1919, while Presi dent Wilson was in France. Hawley (1969) and Bellush (1975) reveal signif icant and acrimo nious policy differ ences between the congressional com mittees, many New Dealers, small busi nesses, and the NRA policymakers. is also one of principals and agents. However, the industrial policy episodes were unusual in certain respects. Government officials found it necessary to assemble the required staff from private-sector resources, and relied little on the existing administrative and per sonnel resources of government. Indeed, the federal government sought and found the necessary expertise among the very indus tries and corporations that would ultimately be subjected to industrial policy regulation. More important, evidence reveals that the preferences of these agents differed from many businesses in regulated industries (especially small business), from the pres ident and congressional committees (princi pals), and from alternative agents (civil servants and military officers employed by other government entities).5 The Wilson and Roosevelt administrations were aware of these important policy differ ences and sought to develop strategies and control procedures that might prevent such differences from distorting public policy. The control procedures involved three elements: instructions, performance monitoring, and incentive systems. haps to the perception of external threats to the nation and the resulting urgency in pol icymaking. The consensus was an illusion. The three agencies only superficially satisfied the diverse interest groups that had supported enactment. With respect to the NRA, Haw ley (1969) comments: As written, the National Industrial Recovery Act could be used to . . . cartelize the economy, estab lish overhead planning, or attempt to eliminate the market riggers and enforce competition. There were those who would move in each [direction] and it was not surprising that a conflict ensued (p. 35). Since the principals—the president, cabi net, and congressional committees—were unclear initially about the role, responsibili ties, and powers of the industrial policy agen cies, they could not transmit clear and intel ligible policy instructions to their agents. As a result, the bureaus themselves became battlegrounds for the definition and details of policy—a divisive phenomenon that delayed the design and execution of policy initiatives. The NRA was plagued by serious internal conflict over powers and responsibilities from the beginning. The conflict interfered with the NRA’s efforts to develop a coherent policy and stabilize the economy. The continuation of wartime crisis and the effectiveness of II. Instructions and government-engineered propaganda muffled Consensus Problems the external criticism of, and, to a lesser The public and political debates that pre extent, the vocal conflict within, the WIB and WPB. Nevertheless, Cuff (1973), Kois ceded the establishment of the WIB, NRA, and WPB occurred in an atmosphere of tinen (1984), and Novick, Anshen, and Truppimpending crisis. Advocates of all persua ner (1949) show that the internal policy sions recognized that a rapid policy response debates of the WIB and WPB were haunted was imperative. In the charged environment, by philosophical conflict. Goldman (1971), Willoughby (1934), Solo (1959), and Garfield legislative and executive actions to create (1921) show that internal contradictions also an industrial policy apparatus were hasty. Congressional debate and media comment of handicapped other wartime agencies, which the day show that the objectives, powers, often delayed or precluded the implementa and policy instruments available to the three tion of important policy initiatives. agencies were not clearly defined. Given the The NRA never seems to have received dramatic expansion of state power sanctioned clear instructions from the Roosevelt admin by the legislation, the absence of controversy istration about the details of policy. The prep seems remarkable, being attributable peraration of the industry codes consumed a tremendous amount of time, and sketchy evi dence of code compliance did not arrive on the president’s desk until the spring of 1934. By Federal Reserve Bank of Cleveland then, the NRA codes were under strong attack from major segments of the business com munity, and many of Roosevelt’s close advi sors had lost faith in the agency’s ability to stabilize the economy. In May 1935, the NRA was declared unconstitutional by the U.S. Supreme Court. Bellush (1975) suggests that the Roosevelt administration adopted a handsoff attitude to the NRA, given the public and legal controversies that surrounded its short, ineffectual existence. In contrast, the agents in charge of the WIB and WPB succeeded in their persistent quest for clearer instructions from Presidents Wilson and Roosevelt, in large part because their planning activities were essential in wartime (in contrast to the NRA). Such instructions usually applied to quite specific regulatory and crisis-related matters, yet at times concerned long-term strategic planning issues. Although the two agencies continued to be viewed by their prin cipals as long-term planning agencies, the force of unpredictable events compelled prin cipals and agents alike to concentrate plan ning resources and regulatory efforts on crisis management. Clear lines of communication and authority between principals, primary agents, and sec ondary agents obviously are necessary for effective monitoring. These communication lines must also be used at frequent inter vals to discuss substantive matters of perfor mance and policy. The historical record sug gests that Presidents Wilson and Roosevelt and their advisors held frequent, substan tive discussions with their wartime economic planning agents. During most of 1917, for example, Wilson (or his close advisors Colonel Edward M. House and Secretary of War New ton D. Baker) met with WIB and CND officials as often as once a week. Such meetings fea tured discussions of WIB powers, relations with other government and military agencies, and Allied purchases in the United States. NRA officials appear to have communicated on a frequent basis with President Roosevelt and his advisors, but the content of discus sion seems to have been oriented, to a greater extent than in the wartime agencies, toward the political repercussions of policy actions. During the first six months of 1919 Presi dent Wilson was absent in Versailles nego tiating the peace, the only identifiable period that frequent communication subsided be III. The Monitoring of tween the principals and agents of industrial Agent Performance policy. It is not mere coincidence that dur The industrial policy principals had to mon ing this period WIB supporters, freed from close supervision, established a peacetime itor the behavior of their implementation price-fixing agency in the Department of Com agents, to assure a minimum of delay, policy merce—the Industrial Board. The board was deviation, and corruption. Principals and dissolved upon Wilson’s return from France, primary agents alike had to select and moni in part because the president believed that tor quantifiable measures of policy perfor wartime powers were not consistent with a mance, to assure that policy was yielding the peacetime economy. desired economic results. This information Communication between the industrial pol could later be used to modify agent instruc icy agencies and participating businesses— tions, to discipline agents who performed in a the primary and secondary agents of policysubstandard fashion, and to reward agents possessed unusual characteristics. Frequent whose performance exceeded expectations. and substantive discussions were typically confined to major defense contractors, large private corporations, and trade associations— Economic Review • Summer 1984 6. The difficulty of measuring the value of government out put to society is not peculiar to industrial policy. Indeed, such problems afflict most government policies. For further discus sion of the prob lems of measuring public-sector output, see Wolf (1979) and Olson (1973). It should also be noted that contem porary theories of marketfailure would not accept the mar ket imperfection argu ments made by NRA proponents. the former and future employers of many toplevel agency staff members. In some indus tries, the WIB, NRA, and WPB officials made few attempts to communicate with the sec ondary agents of industrial policy (especially small and medium-sized businesses), leav ing such efforts to trade association person nel and other industry spokesmen. Such practices exacerbated existing splits within industries, hampering policy execution by reducing the level of secondary agent compli ance with directives. The second aspect of monitoring concerns the ability of principals and agents alike to measure the performance of industrial policy— that is, the effect of policy execution on policy targets. Ideally, planners would have liked to have measured the value to society of margi nal changes in policy targets resulting from policy execution. However, the three indus trial policies were designed to remedy compet- Table 1 An Illustration of Performance Measurements Tiers Policy objectives Final phys ical outputs Final physical inputs Interm edi ate physical inputs WIB Defend the homeland and Allies Enemy casualties Number of rifles and tanks Number of procure ment con tracts Size of strategic stockpile Crude phys Dollars in ical inputs budget Number of employees Number of contracts negotiated 10 NRA Stabilize price and employment levels None Defend the homeland and Allies Enemy casualties Number None of rifles and tanks Number of Number of complaints procure settled ment con tracts Number of Number of signatories product pri ority orders to codes Dollars in Dollars in budget budget Number of Number of employees employees Number of codes w ritten Federal Reserve Bank of Cleveland WPB itive market failures through the use of nonmarket instruments. National defense, a public good, could not be provided in sufficient quan tities by the private sector in World Wars I and II. While there was certainly a public demand for national defense, it possessed no market price, since national defense was (and is) provided by government outside the mar ket mechanism. Without an explicit price, the marginal value to society of government actions to enhance U.S. defense capabili ties, and triumph in war, could not be calcu lated. The NRA also was designed to correct a market failure: New Deal planners and cor porate executives argued that U.S. businesses were too self-centered to recognize that “cut throat competition” was causing a very dam aging deflation in prices and wages. They believed only government sponsorship of explicit collusion might remedy this market imperfection.6 Since government could not calculate the marginal value to society of policy-induced changes in targets, principals and agents had to settle for output or cost measures of perfor mance. Wolf (1979) indicates, however, that nonmarket outputs are usually hard to define in principle, ill-defined in practice, and extremely dif ficult to measure independently of the inputs which produce them. They are generally intermediate products which are, at best, only remote proxies for the ‘real’ or final intended output. Measuring outputs by their inputs becomes accepted because measuring outputs directly is so difficult (p. 113). Table 1 depicts stylized performance mea surement problems that were confronted by principals and agents engaged in the three industrial policy efforts. It is reasonable to assume that proximity to final output and value is associated with superior performance measurement capabilities. Thus, well defined measures of final physical inputs contain more information about final output and value than crude physical inputs. The NRA could not measure the amount of price and employment stability created by an industrial policy designed to affect those economic variables. In the absence of output quantities, the NRA could not derive a mea sure of cost effectiveness in the delivery of output. The New Dealers also had no appro priate measure of final inputs, or the percent of national output, employment, and business firms complying fully with industry codes. The Roosevelt administration and the NRA staff settled for intermediate and crude prox ies in their quest for performance measures, such as the number of industry codes writ ten, business signatories to the codes, and code compliance actions taken and resolved. Yet, these input measures bore little relation to performance. While the NRA staff could count the number of businesses that had signed the codes, they could calculate neither the weight of those businesses in each indus try nor the proportion of signatories actu ally complying with the codes. Likewise, the number of code violations that came to the attention of authorities was a poor perfor mance measure. Bellush (1975) suggests that the largest numbers of code violations came from industries where compliance seems to have been more widespread, while smaller numbers came from industries where little compliance was evident. Like the NRA, the WIB and WPB could not estimate the marginal value to society of their defense mobilization efforts. The mil itary services have sometimes measured their performance (a final physical output of the planning agencies) on the basis of enemy cas u alties; yet, as observers of th e V ietn am con flict noted, casualty rates may not be a satis factory measure of output. The wartime industrial policy agencies had, as their pri mary responsibility, the design and execution of policies that would assure the maintenance of adequate supplies for the war machine. Therefore, performance could be (and was) measured by simply counting the number of physical units such as ordnance, uniforms, medical supplies, and fighter aircraft deli vered to the military establishment. If neces sary, such final physical inputs could be used Economic Review • Summer 1984 to calculate measures of cost effectiveness. In wartime, such measurement may not have mattered (in peacetime, however, military and civilian agencies share measurement problems). Novick, Anshen, and Truppner (1949, p. 16) argue in the war economy, the prices of products needed for the military machine are of no importance. Failure to provide the necessary weapons results in national calamity. . . . It is true that part of the attention of government is devoted to price controls and the avoidance of inflation. Those concerned with this problem are looking out for the general health of the economy during and after the war. Price is never a factor in influencing the satisfaction of the needs of the war machine. Industrial out put for military needs is taken outside the sphere of peacetime economic operations. Although principals and agents shared the problem of performance measurement, the agents possessed an advantage—one that could have been used to exaggerate (inten tionally or unintentionally) their own per formance. The agents selected to implement policy possessed a better understanding of business conditions and industry practices than principals. The historical record con tains no definitive evidence suggesting that the agents employed to implement industrial policy intentionally sought to conceal evi dence of poor performance from their prin cipals. Not surprisingly, the industrial policy agents did argue that the policies they had devised and im p lem en ted w ere responsible for favorable turns of events, but not for unfavorable outcomes. Thus, the moderate expansion of the economy and the relative stability of prices that occurred from 1933 to 1935 was attributed to the NRA. Yet, subse quent analysis has shown that the NRA was not effective in either area. Given their own inexpert knowledge, principals had few ways to judge such assertions on the part of their agents, and therefore were not adequately prepared to issue new instructions to correct the defects in industrial policy design. To measure agent performance, principals had to rely on the observed behavior of agents (cor- ruption, for example, or ability to get along in the bureaucracy) rather than on the effi ciency of agents in carrying out their policy assignments. IV. Rewards and Penalties The relations between industrial policy prin cipals, primary agents, and secondary agents can be guided, in part, by a system of rules promising rewards for performance or behav iors that exceed expectations, and penalties for substandard performance or malfeasance. Information assembled through the monitor ing process ideally is used to operationalize such an incentive system. Two distinct sets of relations and incentive systems occurred in the implementation of industrial policy. Indus trial policy principals (the president and con gressional committees) operated an incentive system as part of their relations with primary agents—individuals employed by the federal government. The policing of behavior and per formance of secondary agents was, in turn, delegated by principals to their primary agents. Businessmen and others consented to serve as primary agents of industrial policy for many of the same reasons that individuals enter government service today—income, job security, status and prestige, amenities, and power. In addition, the government promoted policy as a chance for active participation in a patriotic fight against an external foe. Another incentive may have been the ability to manip ulate the business environment to benefit ones’ own industry. All three agencies con structed elaborate codes of conduct for indus try, regarding output, wages, prices, material costs, collective bargaining, and investment. All three featured price controls as a key ele ment of policy (minimum prices in the NRA, and maximum in the two wars). Cuff (1973) and Koistinen (1980) agree that the codes and price controls may have favored companies and industries that were well represented in the wartime agencies. Likewise, since truly Federal Reserve Bank of Cleveland competitive bidding did not characterize war time procurement procedures, representation may have helped secure military contracts. Finegold and Skocpol (1984) argue that, while business leaders sought to benefit through participation and representation in the NRA, wide divisions in the ranks of secondary agents prevented success. The system of agent incentives constructed by principals suffered from important defects. Although senior agency staff were not offered civil-service protection, the president found it difficult to fire, demote, or transfer agents who exhibited substandard performance or behavior. Agents were protected by their close alliances with powerful private-sector inter ests and congressional leaders, and by the presidents’ inability to assemble clear evi dence of unacceptable behavior or inadequate performance. In addition, many industrial policy agents possessed divided loyalties— divided between their industries and compa nies and the principals by whom they had been hired. Agent actions that, implicitly or explic itly, favored an industry or a company over another were frowned on by principals on the few occasions that they came to light. Yet agents could expect to be rewarded by private-sector interests for such actions. Primary agents were responsible for estab lishing and maintaining an incentive system for secondary industrial policy agents. In wartime, secondary agents were offered such incentives as cost-plus and/or fixed-price con tracts with progress payments prior to deliv ery, a guaranteed market for their output through government procurement, and pro duction priority schemes that legalized sales in the rationed and shortage-ridden civilian economy. Penalties for misbehavior or sub standard performance included the cancella tion of procurement contracts, legal action, and inability to secure high-priority assigna tions. The incentive scheme was not, how ever, administered in an evenhanded man ner. First, primary agents possessed a set of industry-specific contacts that could not encompass all of the companies and produc tion facilities in an industry. Thus, imper fect information about production capacity, prices, and other industry characteristics led to the appearance of prejudice in the admin istration of incentive systems. Second, regula tion by the regulated did not mean that all of the regulated had an equal voice in the main tenance of self-serving policies. Asymme tries in primary agent representation could have biased policy and served to accentuate pre-existing industry divisions (as small and large firms or single- and multi-plant com panies). Third, although incentives existed in regulated civilian markets, industrial policy agencies concentrated their policing in the government-procurement area. The develop ment of a sizable black market during both world wars was one result. Clinard (1969) cites several World War II studies, conducted by the Office of Price Administration, that suggest that 5 percent to 25 percent of civil ian foods and gasoline were sold on the black market in 1943 and 1944. The incentive schemes developed by the WIB, WPB, and their sister agencies were intended to establish “orderly” markets in periods of extremely high demand, by spon soring industry-by-industry collusion and dampening competition. While the schemes may not have been equitable, neither did they seriously retard government-procure ment efforts or transform the entire civilian market into an underground economy. The NRA codes also sought to establish orderly markets, yet the agency was not granted the same degree of legal enforcement power available to the WIB and WPB (even though the wartime agencies’ enforcement powers over the civilian economy were little used). In addition, the encouragement of wartime eco nomic collusion wras intended to benefit, first and foremost, the government and the mil itary machine. In the peacetime NRA, the benefits were supposed to accrue to those who engaged in collusive activities, with a Economic Review • Summer 1984 positive spillover to price and employment levels (objectives of government policy). The NRA could offer few rewards for collusion, and possessed limited ability to enforce disci pline among secondary agents. Moreover, NRA policies served to accentuate existing industry divisions, preventing sufficient numbers of secondary agents from buying into the program. V. Contemporary Industrial Policy Proposals The current proposals for government inter vention in the private economy, under the industrial policy umbrella, contain social and economic objectives quite different from those of the WIB, NRA, and WPB. Never theless, the proposals bear a remarkable resem blance to the historical record. Policy is con ceived as a broad, coordinated, and highly interventionist attempt by government to manipulate the private economy. The policy making and implementing agencies proposed today are structured in a nearly identical manner to the agencies of World Wars I and II and the Depression. The similarities between current proposals and the historical record in these and other areas strongly sug gest that an industrial policy for the 1980s would not escape from the implementation problems experienced in the first half of this century. Initiatives frequently proposed by indus trial policy advocates would involve the fed eral government in a wide variety of private decisions normally handled by individual companies, that is, the most “promising” pro duction technologies to pursue in research and development ventures; the most profit able technologies for production; the optimal location of plants; the most effective methods for withdrawing from declining industries; and the proper degree of corporate diversifica tion. Today’s federal government is larger and employs more expert civil servants than it did four decades ago. Yet, it may not be any better prepared to make such traditionally private decisions, especially in times of peace. The technical and administrative skills nec essary for informed decisionmaking would force the federal government, acting as prin cipal, to seek qualified agents from the pri vate sector. The agents would, in all likelihood, have policy preferences that differ from other inter ested parties, including elected representa tives, alternative agents within federal, state, and local governments, and other industry and labor spokespersons. The primary agents of industrial policy in our three historical cases represented a relatively narrow subset of private businesses in each regulated indus try, although efforts were made to include the voices of organized labor, government, and the public. Organized labor would play a more influential role in the design and imple mentation of an industrial policy for the 1980s. Some proponents believe that labor representation will serve as a counterweight to private business executives. Yet, business and labor sentiments often converge, as evi dent in protectionist trade policies. Such poli cies are certainly detrimental to U.S. consum ers and exporters (see Bryan and Humpage, this Economic Review), but these groups are not likely to be well represented among the agents selected to implement industrial policy. Supportive coalitions for the WIB, NRA, and WPB were forged in times of menacing national crisis. Industrial policy enthusiasts must agree that the U.S. economy is not now facing a similar external threat. The likeli hood of developing a broad coalition in a non threatening peacetime environment must be considered remote. Nevertheless, policy makers have more time to select objectives and design programs that represent a consensus and are potentially effective. Currently, the industrial policy debate remains confused, with no single approach or goal seemingly acceptable to most advocates. Thus, legisla tive action at this point would be premature, leading to unclear instructions for agents. Federal Reserve Bank of Cleveland Contemporary industrial policy proposals seek to increase U.S. productivity and enhance the competitive position of U.S. products in the international marketplace. Like the objec tives of the WIB, NRA, and WPB, the value of policy-induced marginal improvements to society would be indeterminate. Government and its industrial policy agents would face great difficulties in attempts to measure the final physical output of policy. Principals and agents alike would have to settle for crude and intermediate measures of performance that might not have any quantifiable or pre dictable relation to the ultimate goals of policy. The absence of effective performance mea sures will cause difficulties in the design and maintenance of an incentive system for the agents of industrial policy. The application of rewards and penalties might be limited to observable agent behavior (for example, the ability to get along in large bureaucracies) and not related to effective performance in carrying out policy directives. Imperfect information, divided agent loyalties, and asymmetrical agent representation may also make it difficult to gain the trust and com pliance of secondary agents, whatever the formal incentive structure. Good intentions are not sufficient to pro duce effective public policy. If enacted, a coordinated and interventionist U.S. indus trial policy would have to negotiate an arduous obstacle course on the way to delivery. The barriers to effective implementation are not insurmountable: both the WIB and WPB achieved a measure of success. In peacetime, however, the implementation difficulties are likely to be more severe. The arguments and evidence developed in this paper should convince industrial policy advocates and critics that questions of policy implementation are just as important as the content of the pol icies themselves. References Downs, Anthony. Inside Bureaucracy. Bos ton, MA: Little, Brown & Co., 1967. Banfield, Edward C. “Corruption as a Feature of Governmental Organization,” Journal of Fine, Sidney. Laissez-Faire and the General Law and Economics, vol. 18, no. 3 (Decem Welfare State: A Study of Conflict in Amer ber 1975), pp. 587-616. ican Thought, 1865-1901. Ann Arbor, MI: University of Michigan Press, 1956. Batchelder, Ronald W., and Herman Freudenberger. “On the Rational Origins of the Finegold, Kenneth, and Theda Skocpol. “State Modern Centralized State,” Explorations Capacity and Economic Intervention in the Early New Deal,” Political Science Quarterly, in Economic History, vol. 20, no. 1 Ja n u vol. 97, no. 2 (Summer 1982), pp. 255-78. ary 1983), pp. 1-13. Bellush, Bernard. The Failure of the NRA. ______ “State, Party and Industry: From New York: W.W. Norton, 1975. Business Recovery to the Wagner Act in America’s New Deal,” in Charles C. Bright Brand, Donald R. “Corporatism, the NRA, and and Susan F. Harding, Eds., Statemaking the Oil Industry,” Political Science Quar and Social Movements: Essays in History and terly, vol. 98, no. 1 (Spring 1983), pp. 99-118. Theory. Ann Arbor, MI: University of Mich Burton, John. Picking Losers . . .? The Politi igan Press, 1984. cal Economy of Industrial Policy. London: Garfield, H.A. Final Report of the United States Institute of Economic Affairs, 1983. Fuel Administrator, 1917-1919. Wash Chandler, Alfred D., Jr., and Louis Galambos. ington, DC: U.S. Government Printing “The Development of Large-Scale Eco Office, 1921. nomic Organizations in Modern America,” Goldman, Michael Abbot. The War Finance Journal of Economic History, vol. 30, no. 1 Corporation in the Politics of War and Recon (March 1970), pp. 201-17. struction, 1917-1923. Ph.D. dissertation, Christman, Calvin L. “Donald Nelson and the Department of Political Science, Rutgers Army: Personality as a Factor in CivilUniversity, 1971. Military Relations during World War II” Hartz, Louis. The Founding of New Societies: Military Affairs, wol. 37, no. 3 (October 1973), Studies in the History of the United States, pp. 81-3. South Africa, Canada, and Australia. New Civilian Production Administration. Industrial York: Harcourt, Brace & World, 1964. Mobilization for War, Volume I, Program Hawley, Ellis W. The New Deal and the Problem and Administration. Washington, DC: of Monopoly: A Study in Economic Ambiva U.S. Government Printing Office, 1947. lence. Princeton, NJ: Princeton University Clinard, Marshall Barron. The Black Market: Press, 1969. A Study of White Collar Crime. Mont Himmelberg, Robert F. “Business, Antitrust clair, NJ: Patterson Smith, 1969. Policy, and the Industrial Board of the Department of Commerce, 1919,” Business Cuff, Robert D. The War Industries Board: History Review, vol. 42, no. 1 (Spring 1968), Business-Government Relations during World pp. 1-28. War I. Baltimore, MD: Johns Hopkins Uni versity Press, 1973. Industrial Change and Public Policy. Sym Daniels, Belden, and Michael Kieschnick. posium Series, Federal Reserve Bank of Theory and Practice in the Design of Devel Kansas City: Kansas City, MO, 1983. opment Finance Innovations. Washing ton, DC: Council of State Planning Agen cies, 1978. Economic Review • Summer 1984 Koistinen, Paul A.C. The Military-Industrial Complex: A Historical Perspective. New York: Praeger, 1980. ______ “Warfare and Power Relations in America: Mobilizing the World War II Economy,” in James Titus, Ed., The Home Front and War in the Twentieth Century: the American Experience in Comparative Perspective. Proceedings of the Tenth Mili Peirce, William Spangar. Bureaucratic Failure and Public Expenditure. New York: Academic Press, 1981. Pressman, Jeffrey L., and Aaron B. Wildavsky. Implementation. Berkeley, CA: University of California Press, 1973. Rockoff, Hugh. Drastic Measures: A History of Wage and Price Controls in the United States. Cambridge: Cambridge University tary History Symposium, October 20-22, Press, 1984. 1982. Washington, DC: U.S. Government Printing Office, 1984, pp. 91-120. Skowronek, Stephen L. Building a New Amer ican State: The Expansion of National Ad Maunder, Peter, Ed. Government Intervention ministrative Capacities, 1877-1920. Cam in the Developed Economy. London: Croom bridge: Cambridge University Press, 1982. Helm, 1979. Robert A. Synthetic Rubber: A Case Study Mazmanian, Daniel A., and Paul A. Sabatier. Solo, in Technological Development Under Gov Implementation and Public Policy. Glen ernment Direction. Study of the Subcom view, IL: Scott Foresman, 1983. mittee on Patents, Trademarks, and Copy Mitnick, Barry M. The Theory of Agency: The rights of the Committee on the Judiciary. Concept of Fiduciary Rationality and Some U.S. Senate. 85 Cong. 2 sess. Washing Consequences. Ph.D. dissertation, Depart ton, DC: U.S. Government Printing ment of Political Science, University of Penn Office, 1959. sylvania, 1974. Trezise, Philip H. “Industrial Policy is not the ______, and Robert W. Backoff. “The Incen Major Reason for Japan’s Success,” Brook tive Relation in Implementation,” Graduate ings Review, vol. 1, no. 3 (Spring 1983), School of Business, University of Pitts pp. 13-8. burgh, Working Paper Series WP-559, Wescott, Robert F. “U.S. Approaches to Indus August 1983. trial Policy,” in F. Gerald Adams and Law rence R. Klein, Eds., Industrial Policies Nelson, Richard R., Ed. Government and Technical Progress: A Cross-Industry Anal ysis. New York: Pergamon Press, 1982. for Growth and Competitiveness: An Eco nomic Perspective. Lexington, MA: Lexing ton Books, 1982. Novick, David, Melvin Anshen, and W.C. Truppner. Wartime Production Controls. Willoughby, Woodbury. The Capital Issues New York: Columbia University Press, 1949. Committee and the War Finance Corporation. Baltimore, MD: Johns Hopkins University Olson, Mancur. “Evaluating Performance in Press, 1934. the Public Sector,” in Milton Moss, Ed., The Measurement of Economic and Social Wolf, Charles, Jr. “A Theory of Nonmarket Performance. New York: National Bureau Failure: Framework for Implementation Analysis,” Journal of Law and Economics, of Economic Research and Columbia Uni vol. 22, no. 1 (April 1979), pp. 107-39. versity Press, 1973. Pavitt, K., and W. Walker. “Government Poli cies toward Innovation: A Review,” Research Policy, vol. 5, no. 1 January 1976), pp. 11-97. Federal Reserve Bank of Cleveland Both authors are economists with the Federal Reserve Bank of Cleveland. The authors would like to thank Wil liam Gavin, Roger Hinderliter, Ed ward Weber, Paul Gary Wyckoff, and especially K.J. Kowalewski for their valuable comments through out the prepara tion of this article. Diane Mogren pro vided highly com petent research assis tance. Any errors that remain, of course, are the responsibility of the authors. 17 Voluntary Export Restraints: The Cost of Building Walls by Michael E Bryan and Owen E Humpage Before I built a wall I ’d ask to know What I was walling in or walling out, And to whom I was like to give offence. R obert F ro st Economic Review • Summer 1984 Concerned over the impact of foreign com petition, U.S. industries have increasingly turned to elected representatives to seek pro tective barriers against the flow of imports. Calls for protectionist legislation are inten sifying as elections draw near. Proponents of such legislation argue that trade restric tions are necessary to protect U.S. jobs, but protectionist devices usually secure jobs at a substantial cost to consumers and economic efficiency. When building protectionist walls, policymakers should consider the individu als to whom such barriers would likely give offense. The Japanese Voluntary Export Restraint (VER) program, which restricts exports of Japanese cars to the U.S. market, provides a recent example of such a barrier. This article develops a supply and demand equation for new Japanese cars to estimate the price and quantity impacts of the Japa nese VER program. In theory, the impact of a quota, whether voluntary or mandatory, is to raise the price of a good in a given market while reducing sales. Because retail prices for new Japanese cars are not publicly avail able, we must estimate them. The retail price model consists of dealers’ markups over options-adjusted wholesale prices of Japanese cars. The markups vary as dealers adjust their inventory positions to market condi tions. This is an important avenue by which pressures from the VER program influence market prices and consequently market sales. Combined with a demand equation, the whole sale price model is simulated under a set of non-VER assumptions, yielding estimates of the price and quantity impacts of the VER program. These estimates enable us to approx imate measures of income transfers and effi ciency losses associated with the VER program. The results of our study show that the VER program initially had little impact on the new-car market. By 1983, however, the program added over $1,114 to the optionsadjusted price of a new Japanese car, trans ferred roughly $2.0 billion from consumers of new Japanese cars to producers and dealers, and generated $166.4 million in efficiency 1. For a summary of studies on the Japanese cost advan tage in automobile production, see Loos (1984). 2. The quota fig ures cited do not include certain car like vehicles (that is, some four-wheel drive vehicles) that, when included, raise the limitations to 1.76 million units per year in the 1981-83 period and 1.95 million units per year currently. costs. These results are based on a partial equilibrium model that does not consider secondary price effects, that is, the price of substitute cars. While not explicitly considered in this analysis, such effects also could add to the efficiency costs of the program. Sepa rately, we estimate that the VER program “protected” at most 1,500 new jobs for domes tic autoworkers. and quantity restrictions against car imports. Ford Motor Company filed a similar petition in August 1980. The ITC, however, rejected the petitions, finding that imported cars were not an overriding cause of injury to the U.S. car market. Instead, the ITC determined that the U.S. recession and a shift in consumer preferences toward small, fuel-efficient cars were more detrimental to the domestic auto mobile industry than were imported cars. Following the ITC’s decision, the pressures I. The Framework for Analysis to limit car imports were aimed more directly The Setting for Restraint toward the Japanese government. Both the Carter and the Reagan administrations, Until the mid-1970s, sales of intermediatefavoring neither legislated quotas nor tariffs, and full-sized cars dominated the U.S. auto encouraged Japanese voluntarily to limit market. Confronted with rapidly rising gaso their new-cartheexports to the United States. line prices and economic recessions, Ameri Neither administration, however, could rule can consumers dramatically altered their out U.S. legislation, and both the U.S. House automobile preferences in favor of more of Representatives and the Senate introduced economical, fuel-efficient models. By 1980, quota legislation in an effort to pressure the subcompact cars represented the largest com Japanese impose their own limitations. In ponent of the U.S. new-car market, account May 1981,tothe Japanese government agreed ing for 42 percent (compared with 20 per “voluntarily” to cent in 1975 and 12 percent in 1965). Foreign the United States.limit their car shipments to producers, especially the Japanese, had an Japan initially agreed to limit car exports to apparent advantage in the production of the United States over the three-year period small, fuel-efficient cars and gained a sub from April 1981 through March 1984; in stantial share of the U.S. new-car market dur November 1983, the Japanese extended the ing the 1970s.1The Japanese market share agreement through March 1985. In the first rose from 6 percent in 1972 to 12 percent in year, the agreement limited Japanese car 1978. As the decade closed, the U.S. market exports to the United to 1.68 million contracted: domestic new-car sales fell 29 per units, contrasting withStates sales of 1.75 million cent between 1978 and 1980. Japanese sales, units in 1979 and 1.91 million units in 1980.2 however, continued to expand, with the In subsequent years of the program, the VER Japanese market share increasing sharply limitations were to rise by 16.5 percent of to 21 percent by 1980. the growth experienced in U.S. new-car sales As declining domestic new-car sales idled during the previous year. The recession in labor and capacity, the United Auto Workers the United States, however, continued to (UAW) and some of the large domestic car hamper domestic sales: U.S. new-car sales producers began seeking protection from declined from 8.9 million units in the their foreign competitors, especially the Jap actually year preceding the VERs to 8.1 million units anese automakers. In June 1980, the UAW in both 1981 and 1982. Because the U.S. petitioned the International Trade Commis market failed to grow over the first two years sion (ITC), alleging that imports were a sub of the VER period, Japanese limitations stantial cause of serious injury to the domes remained at 1.68 million unitscarthroughout the tic industry and seeking both higher tariffs first three years of the program. Under the current, fourth-year extension of the pro gram, the VER limitations have risen to 1.85 million units. Federal Reserve Bank of Cleveland Theoretical Effects of Trade Restraints Trade restraints limit the flow of foreign goods into domestic markets and raise the prices of imports, the protected goods, and their sub stitutes. In doing so, trade restraints transfer real income away from consumers toward domestic producers of the protected good and foreign producers of the restricted good, and create production and consumption inefficien cies. Figure 1 presents a simple, comparative static model of the U.S. market for Japanese cars and helps to illustrate some of the effects of the VER program. The downward sloping line DjD° in panel A is the U.S. demand curve for imported Japanese cars. Assuming that domestic and foreign cars are close substi tutes, the import-demand curve is constructed as the horizontal difference between the total U.S. demand curve for all cars D jD f and the domestic supply curve SdSfi (see panel B). The import-demand curve DjD° shows the number of Japanese cars that U.S. consum ers would purchase at various prices or, con versely, the maximum per-unit price that U.S. consumers would pay for a given quan tity of imported Japanese cars. The upward sloping line S jS f in panel A is the supply schedule for Japanese cars exported to the United States. It depicts the number of cars that profit-maximizing Japanese producers are willing to export at any given price, or the minimum per-unit price the Japanese produc ers must receive to export a specific quantity of new cars. The intersection of the importdemand curve and the export-supply curve determines the price/quantity combination of new Japanese cars imported into the United States. In the absence of VERs, Q° Japanese cars are sold in the United States at price P°. To illustrate the effects of the VERs, we rely on the concepts of consumers’ surplus and producers’ surplus. For quantities of new Jap anese cars less than Q°, U.S. consumers are Fig. 1 The U.S. New-Car Market Q j ~ Qd + Qj Economic Review • Summer 1984 3. We measure the welfare effects of the VERs following a competitive model. By organizing the Japanese export market, the VER program could confer on Japanese producers increased oligopoly power. This implies a greater welfare loss than measured in our competitive model. 4. The more inelas tic the import de mand curve, the greater the reduc tion in consumers’ surplus resulting from the VER program. 5. The manner in which the Japanese administer the VER program determines how the income transferred from U. S. consumers is divided between Japanese producers and the Japanese government. If, for example, the Jap anese government simply allocates market shares among its pro ducers, all of the income transfer accrues to the Japa nese producers. If, however, the Jap anese government elects to sell the rights to export cars to the U. S. market, the government then will capture some of the additional revenue. 20 willing to pay a price greater than P f as indi cated by the demand curve DjDf. The trian gular area below the demand curve and above the price line, Pfac, represents a benefit that consumers receive called consumers’ surplus. Similarly, Japanese producers are willing to supply quantities of cars smaller than Qf at prices below P f as described by the supply curve SjSf. The triangular area P fcf below the price line but above the supply curve mea sures the producers’ surplus, which accrues to producers as economic rents. The empirical measures developed in this article are related to consumers’ surplus only.3 The use of consumers’ surplus (and producers’ surplus) to measure the costs and benefits of trade policies has generated a great deal of discussion among economists (see Currie, Murphy, and Schmitz 1971). One major aspect of the debate centers on the partial equilibrium nature of the model. Our model considers only the Japanese newcar market, but the VER program could have price effects (upward and downward) on many other goods and services. The prices of new cars, domestic and imported from other foreign countries, for example, probably will rise following the VER program. Usedcar prices also could rise. Any increase or decrease in consumers’ surplus associated with such secondary price effects is relevant to the calculations of the costs and benefits associated with trade restraints. By consider ing only the Japanese new-car market and assuming that the price impacts in other car markets are small, we have understated the income transfers and efficiency losses associated with the VER program. Many economists also argue that consumers’ sur plus should be measured under demand curves that are compensated for the income effects of price changes. We have not done so under the assumption that the income effects associated with the VER-induced price increases are small (see Willig 1976). Despite these and other concerns about the use of partial equilibrium models and the concept of consumers’ surplus for measuring the Federal Reserve Bank of Cleveland effects of trade restraints, the approach has been widely used (see Morkre and Tarr 1980). Our study views the VER program basi cally as a quota organized by the Japanese government (see Bergsten 1975). If the Japa nese government imposes a quota equal to Qj in figure 1, panel A, the U.S. price of Japanese cars will rise to Pj, since this is the price U.S. consumers are willing to pay for Qj Japanese cars. The quota effectively shifts the Japa nese supply schedule from the diagonal SjSf to feS'j, which becomes vertical at point e. As the price of new Japanese cars rises, some U.S. consumers will switch to new domestic cars. The price of domestic cars will rise and production will increase Q j- Q§, as shown in figure 1, panel B. As the prices of new Jap anese and American cars rise, fewer cars in total will be produced and purchased. By raising the price of imported Japanese cars to Pj, the VER program reduces the con sumers’ surplus in panel A by an amount given by area PfPjbc.4 Part of this reduction in consumers’ surplus represents an income transfer to Japanese producers from U.S. con sumers who continue to buy Qj imported Jap anese cars at PJ - P f more per car than before the VERs. This income transfer, shown by area PfPjbd, does not represent a loss to the world economy, but it does represent a loss to the U.S. economy, especially in the short run.5 Although most of the income transferred from U.S. consumers to Japanese producers eventually returns to the United States as foreigners buy U.S. exports and invest in the United States, such transactions could take many years to complete. Even in the long run, the United States could incur a loss if the price of U.S. imports rises relative to the price of U.S. exports because of the VERs. The second part of the overall reduction in consumers’ surplus represents a net loss to both the U.S. and world economies and is given by area bed in figure 1, panel A. Because we constructed the import-demand curve in panel A as the horizontal differ ence between the total-demand curve and the 6. The more inelas tic the Japanese sup ply curve, the greater the loss in producers ’ surplus resulting from the VER program. domestic-supply curve in panel B, the loss depicted by area bed in panel A under the import-demand curve represents the combined losses given by areas abc and def in panel B. The first of these losses results from greater inefficiency in car production. An increase in domestic production of Qd - in panel B replaces part of the reduction in Japanese new:car imports. Before the VERs, the Japa nese provided these additional units at a total cost of cbQdQd, but domestic producers can provide these units only at an additional cost given by area abc. This area measures pro duction inefficiencies resulting from the VER program. The second net loss area, def, is a loss in consumers’ surplus. Because of the VER program, total car purchases have fallen from Q f to Q j , and consumers incur a loss represented by the area def. This area repre sents the consumers’ surplus that would have been received from buying these units at a price below their demand schedule (except at the margin). From an international perspective, area dee (figure 1, panel A) also represents a net loss measured in terms of foregone producers’ sur plus.6 Japanese producers intially bear this loss. Because this article focuses on the costs of the VER program to the United States, and because we lack sufficient information about Japanese production costs, we do not measure this loss. Nor can we estimate the change in Japanese producers’ profits resulting from the VER program. However, we can approximate the change in Japanese producers’ net reve nues via panel A. Area PfPjbd represents addi tional revenues from higher prices on the Qj units sold after the VERs are imposed. Area dcQJQ'j measures the revenue lost because of a reduction in total units sold. The differ ence between these two areas is the net effect on revenues, and it can be positive or nega tive, depending on the price elasticity of the import-demand curve. If the demand curve Economic Review • Summer 1984 is elastic, Japanese producers will lose reve nue as a result of the VER-induced price increases. This revenue estimate provides a link between our model of Japanese new cars and domestic new cars. In section II of this article, we develop an econometric model to estimate the effects of the VER program on prices of new Japanese cars and on the number of new Japanese cars sold in the United States. Following the theo retical analysis in figure 1, panel A, we can then obtain estimates of the income transfers and efficiency losses associated with the rise in Japanese new-car prices and borne by U.S. consumers. These measures are as follows: (1) transfers from U.S. consumers of Japanese cars to Japanese producers = (Pj-Pf)Q j; (2) net efficiency losses = 1/2 ( P j - P f ) ( Q J - Qj). Similarly, we can measure the loss to Japanese new-car manufacturers as (3) change in Japanese producers’ revenue = [ ( P j - P f ) Q j ] - [P°(Q°- Qj)]. The empirical model departs from the theoretical model presented in figure 1, as it introduces into the analysis the role of U.S. dealers of Japanese cars. (A theoretical discussion of dealers appears in section II.) The measures of the consumers’ surplus loss described in figure 1 remain basically the same except that we calculate the income transfers to Japanese producers using the change in wholesale prices; we estimate the income transfer to U.S. dealers of Japanese cars using the change in the dollar value of dealers’ margins, and we calculate the effi ciency loss borne by U.S. consumers using the change in the transactions price of new Japanese cars. With the introduction of U.S. automobile dealers into the analysis, we also broaden the measure of revenue lost to include both Japanese producers and U.S. dealers. 7. See Santoni and Van Cott (1980), Falvey (1979), and Feenstra (1982) for references and examples. 8. When the re straint is based on value, such as an ad valorum tariff, the incentive to upgrade quality does not exist. Quality increases raise the value of the product and would raise the import duty by a proportional amount under an ad valorum tariff. There is no advan tage in this case to improving quality. 9. In contrast, a restriction based on the value of imports such as an ad valo rum tariff would raise the price of each car by the tariff rate, 6, Pi= Pi (1 + 0), and Pi = P2( 1 + 0). Relative prices would not change: Pi = P2(l + e) P{ PxiX + e) ' 22 The Quality Adjustm ent Phenomenon quality of imported Japanese cars should increase in response to the VERs.8 Before turning to the empirical section, we second explanation of quality upgrad should consider the possibility that the VER ingThe involves a shift to a better grade of mer program could alter the quality composition chandise within a broad product category, of Japanese cars. If quality improvements largely reflecting to restraints occur, measures of the income transfers and from the demand adjustments side of the market (see Fal efficiency losses associated with the VERs vey 1979). Consider a Japanese car market will be overstated since price changes will of two models that are close, but reflect greater product services in addition to consisting not perfect, substitutes. The models differ in economic rents. This section considers theo terms of quality and options, and their prices retical arguments for believing that quality reflect these differences. Following Feenstra upgrading occurs as a result of the VERs. (1982), the basic and higher-quality Japa When faced with trade restraints on im nese cars will have unit production of port quantities, foreign manufacturers often Ci and C2 , respectively, such that C\costs is less upgrade the quality of their products in an C2 . In the long-run competitive equilib attempt to maintain profits in the restricted than rium with no restraints, the following market. The effect has been observed in condition holds: the markets for imported textiles, footwear, dairy products, steel, and, recently, Japanese (a) P\ = Ci < P2 - C2 , cars.7 Quotas, specific tariffs, and VERs are where Pi and P2 are the respective car prices. examples of such quality-altering trade re a VER program, the prices of both cars straints. Trade restraints based on the value Under will rise to P{ and Pi, respectively. At the of imports do not promote quality upgrad margin the manufacturer ing. The theoretical literature offers two expla will produceprofit-maximizing the car yielding the nations for the quality-upgrading phenome return, ensuring that at the new highest equilibrium non (see Feenstra 1982). (b) P{ - Ci = Pi - C2. One explanation of quality upgrading involves changes in the specifications of a single, narrowly defined commodity that pri Substituting from equation 1 yields (c) P{ - P, = Pi - P2. marily reflects the response of foreign pro ducers to trade restraints (see Rodriguez 1979 The profit-maximizing producer will raise the and Santoni and Van Cott 1980). According prices of both cars by the same dollar amount, to this view, an imported good consists of a but this implies a reduction in the price of bundle of characteristics appealing to con the more expensive car compared with the sumers. A car, for example, provides trans price of the less expensive car:9 portation, comfort, and aesthetic appeal. A VER limits the physical quantity of an A imported good, not the composition and Pi < Pi amounts of other characteristics embodied in Under some rather restrictive assumptions the product. Foreign suppliers, facing quan about the own-price and cross-price elas tity restraints, have a strong incentive to upgrade and increase the unrestrained attri ticities for the car models, a larger proportion of the expensive cars will be sold (see Falvey butes of their product. In the car example, VERs limit the amount of transportation that 1979, pp. 1106-8). Even though the VERs Japanese producers can sell in the U.S. mar reduce the overall quantity of new Japanese ket, but not the amounts of comfort and aesthetic appeal that they provide. The Federal Reserve Bank of Cleveland 10. This figure is a rough estimate. In any given year, the number of vehicles actually arriving in the United States need not equal the number of vehicles exported from fapan because of a ship ping time lag. A small number of vehicles not covered by the 1.68-million unit limit (for ex ample, some fourwheel drive vehicles, which have a sep arate quota) are included in sales data. The VERs do not include Japa nese cars shipped to the United States via another country, which may also cause a slight data discrepancy. cars, the proportion of more luxurious car imports will rise. The income transfers and efficiency losses associated with VERs when quality upgrad ing occurs are less than those associated with VERs when quality remains unchanged (see Feenstra 1982, p. 14). When quality upgrad ing occurs following the imposition of trade restraints, the observed price increases in clude both the rents resulting from artificial restraints in the marketplace and the costs of the improved quality. In measuring the welfare effects of VERs, price increases attributable solely to quality improvements in the restricted goods can be excluded, as these do not reduce consumers’ economic well-being. II. The Model We estimated a supply and demand model for the Japanese new-car market in the United States. Unfortunately, data on retail (transac tions) prices of new Japanese cars, necessary to estimate the two equations, are not avail able. As did earlier researchers, we found our selves in the unenviable position of having to construct a measure of transactions prices from available wholesale price data. In this section, we envision a price-setting process whereby new-car transactions prices consist of three components: wholesale cost, constant markup (reflecting overhead costs), and variable markup (reflecting temporary shortage or surplus in the marketplace). In addition to the wholesale price, we assume that the variable markup is a major chan nel of transmitting VER influence into the marketplace. The signal by which U.S. deal ers recognize changing market conditions, and consequently adjust variable margins, is provided by dealers’ inventory pressures. In our analysis we first built an inventory model of new Japanese cars from which we derived a ratio of desired to actual Japanese new-car inventories. This ratio is the crucial (and admittedly tenuous) link to the variable Economic Review • Summer 1984 margin component. Next, we formulated a wholesale cost equation, where the dealer cost of cars is a function of unit sales, the bundle of car options, and the dollar/yen ex change rate. Using a two-stage least squares regression technique, we estimated models of wholesale prices and Japanese new-car demand with quarterly data over an eightyear period (1976:IQ to 1984:IQ). The Basic Framework with Inventories The quotas on new cars from Japan restrict imports rather than sales. Given the existence of inventories, sales of Japanese cars need not directly equal imports. (1) Japanese sales = imports from Japan - change in inven tory stocks. As a result, the degree to which quotas bind the marketplace and generate consumer price increases depends on the state of inventories over the period. In 1981, sales of new Japa nese autos (and utility vehicles) exceeded the quota by 21,000 units.10 Over the same period, inventories of new Japanese cars fell 34,000 units. Did the first year of the quotas influ ence Japanese new-car prices? Only insofar as the quotas forced dealers’ inventories below desired levels, which respond to overall eco nomic conditions. Just prior to the VER program, for example, U.S. interest rates reached unusually high levels and the auto market was in a state of cyclical contrac tion. U.S. dealers of Japanese cars could have intentionally liquidated inventory stocks in response to the declining market environ ment. If the inventory corrections that oc curred in 1981 were adjustments to desired levels and not quota-induced, the VER pro gram would be nonbinding, having no price, quantity, or related effects. The initial task, therefore, is to measure actual Japanese newcar inventories relative to a desired inven tory position. Virtually all new-car inventories are held by independently owned dealerships that compete in a highly competitive marketplace where inventory stocks and retail prices adjust rapidly to market conditions. If an overstock of inventories exists, new-car deal ers will lower prices, thus stimulating sales; if inventories fall below desired levels, dealers increase retail prices. Such a market-clearing process has important implications to our analysis, implying that the transactions prices of new cars are tied to dealers’ inven tory positions and that transactions price patterns can deviate from wholesale newcar price patterns. Inventories and variable dealers’ margins in the market-clearing process allow the deal ers’ supply curve, S^(the effective short-run market supply curve), to deviate from the manufacturers’ supply curve, Sm, over short periods of time (see figure 2). To illustrate, imagine that foreign manufacturers are in a state of equilibrium in that they import an amount equal to consumer demand into the domestic market. Such a state could be rep resented by point A in figure 2. The dealer is in a state of equilibrium in that inventory stocks are at desired levels. In such a case, the retail price of the product, P \ , is equal to the wholesale price plus a constant dealers’ margin. Suppose that demand in the market place suddenly surges from D\ to D 2 . Inasmuch as increased foreign imports require a sub stantial lead time for delivery from a foreign manufacturer, sales in part will be main tained by an “undesired” inventory liqui dation at the dealers’ level: the dealers expe rience inventory shortages. Dealers are induced into the shortage condition because of higher margins (or dealers’ rents) as con sumers bid up the existing price of the product in short supply (represented as the difference between P\ and P 2). A short-term equilibrium is achieved at point B in figure 2, where the dealers’ supply curve intersects the market demand curve. However, this is only a temporary equilibrium, since compe tition and manufacturers’ pressure will tend to increase dealers’ supplies. Shipments are made by the foreign manufacturer, and Federal Reserve Bank of Cleveland dealers restock inventories to desired levels. Consequently, the dealers’ supply curve trav els along the manufacturers’ supply curve (S^to Sd) as the level of imports is increased, and a more lasting equilibrium is reached at point C. Here, dealers are again earning no economic rents, while manufacturers are selling more of the foreign product at higher wholesale prices. Retail prices in the final equilibrium, P3, are higher than the initial retail price, P\, but lower than the retail price immediately following the demand shift, P2 . The existence of inventories can dramati cally change the standard supply-constraint analysis shown in figure 1 to an environment more accurately resembling that in figure 3. Notice that, in situations where the existence of inventories is significant, the supply curve of the restricted product does not become vertical until inventories are zero. That is, the point at which the market supply curve becomes fully binding, Si, is lengthened by the amount of the inventory stock, Si to S 2 . Indeed, the temptation is to argue— erroneously—that import restrictions need never be binding as long as inventories exist. Imagine that in the current period we are at equilibrium with respect to inventory stocks, or the dealers’ supply curve intersects the manufacturers’ supply curve at current unit sales levels, represented by point A in fig ure 4. Notice that this equilibrium sales quantity is also equal to the level of imports from the manufacturer, since only if unit sales match imports will actual inventories equal desired inventories. Next we institute a limit on imports such that they cannot exceed the level designated by the current unit sales level. If demand increases such that unit sales exceed the quota, undesired inventory liquidations occur, and dealers earn rents (fig ure 4, point B). Inasmuch as the quota pre vents further shipments from a foreign coun try, the dealers’ supply schedule is essentially fixed to th e in tersectio n of th e m a n u fa c tu r converges on th e im p o rt q u o ta c o n stra in t, the e rs ’ su pp ly cu rv e at point A . O ver tim e, as d e a le rs’ su pp ly cu rv e becom es m ore vertical, inv en to ries are depleted and app ro ach zero, and, in th e lim it, th e d e a le rs’ su pp ly cu rv e th e v ertical su pp ly c o n stra in t, S 2 , collapses becom es v ertical at th e q u o ta level. E v e n tu leftw ard to w ard th e im p o rt q u ota, S\ (figure 5). ally, th e m ark et w ill settle in a quota-induced eq u ilib riu m at point D. L ikew ise, as th e v ertical su pp ly c o n stra in t Fig. 2 Inventories and Supply Fig. 3 Inventories and VERs Prices Prices Units Fig. 4 Short-run Equilibrium Prices Quota plus inventories 1 1 Units Prices Si Economic Review • Summer 1984 Im port quota Fig. 5 Long-run Equilibrium Units S, Units 11. See appendix fo r a detailed descrip tion of the variables used in this article. 12. Although many alternative fu n c tions were esti mated, all yielded strikingly sim ilar results. 26 The Inventory Model Irvine (1979) has demonstrated that monthly models of dealers’ new-car inventory behav ior outperform similarly specified quarterly models. Inventory adjustments at the deal ers’ level proceed rapidly, and, consequently, adjustments between desired and actual deal ers’ inventory holdings often occur within a matter of months. That is, inventory dis equilibrium in the absence of constraints does not persist over long time horizons. For this reason, we have chosen to specify a Feldstein/Auerbach target-adjustment inventory model that allows rapid inventory adjust ment to a relatively slowly changing, desired inventory target (Feldstein 1976). This inven tory model hypothesizes that dealers’ inven tories are equal to a desired inventory posi tion plus a fraction of the unanticipated dealers’ sales forecast error and a random error (see Feldstein 1976, p. 369). (2) It = I f + d(Set - St) + where It = actual unit inventories in period t, I f = desired unit inventories in period t, St = actual sales in period t, Set = dealers’ sales expectations in period t.u Dealers hold inventories for many differ ent reasons. Primarily, dealers hold a desired level of inventories to provide an immediate supply to the retail market and to meet a con tinuous demand for sales between periodic deliveries from a foreign manufacturer. These inventories must further allow for a suffi cient variety of models to satisfy alternative consumer preferences. As sales increase, dealers’ inventory holdings should increase. Inventories also involve costs to the dealers in terms of wholesale price and borrowing expenses. These costs tend to offset the sales influence on the dealers’ desired inventory Federal Reserve Bank of Cleveland positions. Therefore, following Irvine (1979, p. 3), we specify a dealer’s desired inventory position as a linear function of sales and the real cost of holding inventories, or (3) I f = aiSf + a2StCt, where Ct = real marginal inventory holding cost during the current quarter. Substituting equation 3 into equation 2 and dividing through by sales, we can estimate new-car inventory behavior in quarters-supply form. A nonseasonal, quarterly model was specified to isolate the influence of seasonal fluctuations in new-car inventories that eventually translate into seasonally sensitive transactions prices.12 The equation was esti mated over the non-VER period, 1975:IQ to 1981:IQ, using second-order serial correla tion correction. The /-statistics are in paren theses. The estimation horizon does not include the VER period, since we expect that the quotas may have artificially prevented actual new-car inventories from maintaining a “desired” level and consequently precipi tated a long-term inventory shortage. Since we do not have an accurate measure of the VER influence on inventories, estimating a Feldstein-Auerbach inventory specification over the full time horizon runs the risk of biasing the regression results. The estimated inventory equation was It/St = 1.51 - 0.055Q + 0.132(Sf - St)/St (3.27) (0.72) - 0.23DQ2 - 0.31DQ3 + 0.08DQA (3.63) (3.80) (1.32) + 1.217RH01 - 0.753RH02, (9.75) (3.50) (D W = 1.90, R2 = 0.72, SEE = 0.162, F = 9.78). 13. After adjusting for the first-quarter seasonal dummy, the constant value is 1.39. 14. The presence of a strong seasonal pattern can also account for some of the serial correla tion detected in this equation, inasmuch as these dummies are only a crude approximation of what may be a much more complex sea sonal influence. The results of this model were encouraging in coefficient values and fit, but discourag ing because of the presence of second-order serial correlation. Many factors can generate serial correlation. Our primary concern is that we may have omitted a determinant of new-car inventory behavior. One possible omission, found in Irvine (1979), is a price expectation variable that enters into the inventory cost component. Dealers can hold inventories in anticipation of future price increases that would yield capital gains on new-car stocks. Attempts to capture such a “speculative” auto inventory behavior using various expected inflation measures were all unsuccessful. However, the existence of serial correlation may simply be a consequence of a desired inventory level that displays a ten dency toward inertia. Indeed, the significant, but small, value on the cost variable sug gests that target inventory levels adjust quite slowly to changes in inventory-holding costs. The constant term is significantly larger than 1, which implies that dealers maintain an inventory buffer against random fluc tuations in sales.13 The high coefficient also Fig. 6 Desired Dealer Inventories Relative to Actual Ratio 27 1 Economic Review • Summer 1984 Years 2 3 could reflect holdings of a mix of model types. The presence of large and highly significant dummy coefficients (DQ2 , DQ3, DQ4) indi cates a strong seasonal pattern in the auto mobile market. Relative to sales, auto inven tories are most plentiful during the first half of the model year (IVQ and IQ), while inven tories relative to sales dramatically decline during the third quarter as manufacturers make the transition to new models.14 Al though further research may result in supe rior inventory specifications, the econometric model presented here is consistent with pre vious auto inventory analysis. From these estimates, we calculated the dealers’ desired inventory-to-sales position. The desired dealers’ inventory-sales ratio can be expressed as a ratio of total inventories to sales, yielding a quarterly measure of newcar inventory pressure called SHORT (that is, SHORT = /*//). As the ratio increases (or decreases) from 1, a shortage (or surplus) of inventories develops at the retail level. Fig ure 6 shows the estimated quarterly fluctua tions of desired Japanese dealers’ inventories relative to actual inventories over the 1975:IQ to 1984:IQ period. In the pre-VER period, the series accurately illustrates quarters that experienced sub stantial inventory shortages or surpluses. For example, between the second and third quar ters of 1975, a shortage of domestic subcom pacts, combined with surging gasoline prices, fueled a mini-sales boom in the Japanese new-car market. Consequently, actual deal ers’ inventory positions temporarily strayed from desired levels. After showing equilib rium in inventories for the earlier quarter, the shortage variable rose to a value of 1.74 in 1975:IIIQ, or desired dealers’ invento ries of Japanese cars were approximately 74 percent above actual levels. The index fell to 1.23 in the following quarter. In mid-1977, an unexpected, record-producing surge in Jap anese new-car sales created another inven tory shortage: the shortage variable rose from 1.01 in 1977:IQ to 1.70 in 1977:IIQ to 2.06 in 1977:IIIQ. By 1977:IVQ, the shortage virtually 15. The equilib rium, or trend, markup was calcu lated using annual data found in Con sumer Reports (April issues, 1977 through 1983). The markup represents the percentage dif ference between dealers ’ costs and list prices of a salesweighted composite of Japanese models. 28 disappeared, and the ratio of desired inven tories relative to actual inventories remained at 1.17. Using the non-VER inventory model, we calculated inventory shortages over the VER period, attributing any inventory shortage to the effects of the VER program. These esti mates suggest that a meaningful shortage of Japanese cars at the retail level did not exist during the VERs’ first year, as the begin ning of the program coincided with a cyclical decline in Japanese new-car sales and rela tively high U.S. interest rates that induced dealers to liquidate their desired inventories. According to our estimates, an overstock of Japanese new-car inventories existed at the dealers’ level, on average, throughout the first year of the VER program. Although more binding during their second year, the VERs were weakened as the market for new Japanese cars continued to deteriorate during 1982. Inventories deviated substantially from desired levels by year-end 1982, however, and during the third year of the VERs the Japanese new-car market experienced unprec edented shortages. atile and wholesale prices are not), a constant margin model may fail to reflect the underly ing price mechanism. We have augmented the constant margin model to allow for two deal ers’ margin components: (1) a long-run mar gin (or marginal costs at the dealers’ level), and (2) a short-run margin (or the dealers’ rents), as shown in equation 4. (4) Pj= V- M - R, where Pj = transactions price of new Japa nese cars, V = wholesale value of new Japanese cars (in dollars), M = 1 + marginal cost margin, R = 1 + dealers’ rents. The long-run equilibrium margin, M, was con structed by using 1 plus the average annual markup of Japanese cars (dealers’ cost vs. list price); over the 1977 to 1984 period, this mark up averaged 16.8 percent.15 The short-run margin, R, was assumed to be a function of short-run market conditions, or the relative shortage of dealers’ inventories, as speci fied in equation 5: The Price Model (5) R = (SHORT,.!)1*, Data on the dealers’ (wholesale) costs of where new Japanese cars are readily available, but k = price adjustment parameter that there are no publicly available data on Japa determines degree to which inven nese new-car transactions prices. We have tory disequilibrium generates attempted to specify the relationship that transactions price changes. ties these two price series together. The link between the wholesale cost to dealers and the This specification is appealing because market price ultimately paid by consumers it allows undesired fluctuations in invento depends on dealers’ margins. As a transac ries to translate into the price variable, as tions price proxy, earlier analyses have used described in figure 2. We maintain the wholesale price data (dealers’ cost of new competitive property of no long-run deal cars) adjusted by a constant dealers’ margin ers’ rents, since desired dealers’ inventories (see Feenstra 1982; Carlson 1978; and Carl tend toward actual inventories over time son and Umble 1983). However, in an environ (in the absence of supply constraints). As a ment where dealers play an important role in matter of empirical convenience, this specifi the market-clearing process (that is, under cation allows us to substitute the inventory conditions where inventories are highly vol shortage variable into the demand equation as a proxy for fluctuations in dealers’ mar gins—the missing link in the transactions price specification. Federal Reserve Bank of Cleveland This is not to say that dealers are assigned all of the market-clearing responsibility. In deed, there exists a manufacturers’ supply function, which makes it necessary to esti mate a wholesale price equation that is responsive to market conditions. As a result, our wholesale price model includes a sales variable in an effort to capture the influence of market demand on Japanese new-car prices. Also embedded in the wholesale price data are changes in product quality. As discussed in section I, if product quality changes are significant, studies that fail to adjust for the influence of these changes on the price var iable will necessarily bias the estimated price elasticity. As a proxy for such quality shifts, we constructed an index of five Japanese newcar options (automatic transmissions, power steering, air conditioning, sunroof, and audio systems), using data from Ward’s Automo tive Reports (see appendix). The historical pat tern of the Japanese options index is shown in figure 7. The Japanese options index has risen con siderably since its 1975 base year, with a strong underlying trend. It is likely that the trend increase in the quality index simply reflects a maturing product life cycle. The Fig. 7 Japanese Options Index Index 1975:IVQ = 100 Japanese entered the new-car market as a low-margin, high-volume competitor. Having succeeded in this market, the Japanese sub sequently expanded the boundaries of their market influence. After 1982, the second year of the VER program, the options index rose noticeably above its historic trend rate of increase, suggesting further that the VER program had little impact in the market place during its initial years. Beyond 1982, as inventories of new Japanese cars sharply tightened, the options index rose substan tially faster than trend. Our wholesale price model also accounts for fluctuations in the dollar/yen exchange rate, which influences the wholesale price of Japanese cars to U.S. dealers. That is, as the dollar/yen exchange rate rises, the dollar wholesale price of new Japanese cars will rise, other things being equal. The resulting wholesale price model is shown in equation 6: (6) VJ72t = f(SALESt, OPTIONt, YENt), where VJ72t = real wholesale cost of new Japanese cars, SALESt = per capita Japanese newcar sales, in units, OPTIONt - Japanese options index, YENt = dollar/yen exchange rate. The Demand Model The final step in our modeling process in volves the specification of the demand equa tion. Since Chow’s seminal work in 1957, auto demand models have been cast in a stockadjustment framework, where sales are a function of the difference between desired and actual auto stocks (called new demand) and auto stocks of the last period (called replace 240 1976 1978 1980 1982 Economic Review • Summer 1984 1984 16. The two-stage estimation proce dure used in this analysis is the Fair estimator with serially correlated errors. The compu ter program we used was PEC 9.1 (Jon K. Peck, PEC (Program for Econ ometric Computa tion), Version 9.1, Yale University, March 1982). Care should be used when interpreting the t-statistics, as the standard errors are not exactly cor rect. For a thorough discussion of the procedure, see Fair (1970). 17. An alternative procedure would have been to esti mate the equation over the non-VER period and fore cast out of sample through 1984: IQ. Because of the small sample problems that already exist, this alternative did not appear feasible. ment demand). Desired stocks are determined generally by relative prices, real income, and demographics. (7) St = f(K *~ K t_i) + d(Kt.i), where f and d - functional forms, s t = sales of new cars in cur rent period, Kt. i = stock of cars in previ ous period, K f = desired stock of new cars in current period. and (8) K t = f(P t, G^4S72/, Y72t,HHt), where Pt - price of new cars relative to rate of inflation, GAS72t = weighted operating costs of cars relative to rate of inflation, Y72t = real per capita perma nent income, HHt - U.S. civilian, noninstitutional population. Unfortunately, the stock-adjustment approach does not lend itself easily to this analysis, as data on auto stocks are available on an annual basis and therefore limit our ability to estimate the impacts of an importrestraint program that has been in effect since April 1981. As in many auto-demand studies, we instead chose to estimate directly per capita new-auto sales as a function of desired stock explanatory variables: real price (specified in equation 4), real operating cost, and real income. (9) SALESt = f(P]t, GAS72,, Y72t), where SALESt = St/HHt, Pjt = relative price of new Japanese cars. 30 Federal Reserve Bank of Cleveland Estimation Given the wholesale price model of equa tion 6, and having specified the transactions price mechanism in equation 4, we can esti mate the supply and demand equations for the Japanese new-car market. The equations were estimated using two-stage least squares with first- and second-order serial correla tion correction.16 The estimation period was 1976:IQ to 1984:IQ.17 (10) LVJ721 = 4.63 + O.IOLSAL^S, (2.24) + 019LOPTIONt + 0.27LYE Nt (4.46) (2.42) + 0.96RH01 - 0.32RH02, (5.82) (1.96) ( DW= 2.21, R 2 = 0.98, SEE = 0.027, F = 590.0). (11) LSALES, = + + -4.84 - 1.31LF/72, (1.17) 0.\7LSHORTt_x (2.07) 0.60LGAS72t (1.26) + 7.54LF72, (2.15) 0.53RHOI - 0.32RH02, (3.21) (1.95) ( DW= 1.92, R 2 = 0.65, SEE = 0.132, F = 13.2). where L - SHORTt_! log form, = iU / It-i. The coefficient on the log-adjusted wholesaleprice variable (LVJ72) is a direct estimate of the own-price elasticity of new Japanese cars, -1.31; the coefficient on the log markup var iable (LSHORT) represents this price elastic 18. Remembering that the price specification is Pj = V- M - R, the estimated price elas ticity is the coeffi cient on the log of Pj, or, in this instance, /30lnPj = p 0(\nV + In M + k\nSHORT). Our estimate of k, the sensitivity of Japa nese inventory shortages on trans actions prices, is 0.133. (An estimate of 0 suggests that inventory shortages play no role in the price-setting process, while an estimate of 1 suggests that per centage changes in shortages generate equal percentage changes in transac tions prices.) ity times the dealers’ rents adjustment, k.m Our price elasticity estimate is slightly larger than that found by Carlson (1978) for subcompact and compact cars and slightly smaller than that found by Carlson and Umble (1983) for compact cars.19 The other components of the Japanese newcar demand model conform to our expecta tions and to the results of other auto-demand studies. As nearly all auto-demand studies find, real permanent income is the primary determinant of new-auto sales. The Japanese market is no exception. In addition, the cost of operating cars has a positive influence on Japanese new-car purchases. Carlson (1978), Carlson and Umble (1983), and Tishler (1982) had similar findings for the subcompact and compact new-car markets. III. The Results Model Simulations Our final task is to simulate the wholesale price and demand equations under a set of non-VER assumptions to speculate on the Fig. 8 Japanese New-Car Prices quantity and price combinations that would have resulted in the absence of the program. These simulations are highly dependent on our assumptions concerning non-VER market conditions and thus should be viewed with a measure of caution. However, each assump tion was made to allow for the maximum price and quantity impacts on the system. In the price equation, we assume that, in the absence of the VERs, the options index would have maintained its 1981:IVQ value through out the 12-quarter VER period.20 In the demand equation, we assume that dealers’ inventory positions would have maintained an equilib rium ratio of 1 over each of the VER years.21 The results of these simulations are shown in table 1. According to our simulations, in its first year the VER program had virtually no effect on the U.S. automobile market. The options-adjusted transactions prices of new Japanese cars increased by only $11 per unit because of the restrictions, primarily reflect ing a rise in wholesale prices. Dealers did not significantly raise their markups following the VER program; they experienced an over stocked inventory position prior to the VER program that lasted halfway through the pro gram’s first year. The VERs consequently Thousands of dollars 10 1976 31 1978 1980 Variable m arkup 1982 Constant m arkup Economic Review • Summer 1984 1984 Although we use real, options-adjusted prices through out the estim ations, nominal unadjusted price esti mates have been included in figure 8 to show the be havior of actual consumer prices over the VER period and to compare these w ith the price estim ates found in a constant m arkup model. The variable margin specification yields an average price that seasonally fluctuates around the constant m arkup price vari able throughout the non-VER period. As VER-related shortages develop in the marketplace though, the constant m arkup price m easure fails to register meaningful increases compared w ith the dram atic acceleration in prices estim ated by our approach. For example, our estimates suggest that the average price of a new Japanese car during 1983 was approximately $8,315, while a constant markup approach would yield an average price of $7,385. 19. Carlson (1978) found a price elas ticity of 0.82 for the subcompact newcar market and 1.21 for the compact mar ket. Carlson and Umble (1983) found the price variable to be insignificant in the subcompact market, with a compact new-car price elasticity of 1.47. 20. In fact, it is quite likely that the Japanese would have continued some options up grading in light of such a strong trend component. This result would tend to decrease the influ ence of the VERs in the simulations. 21. Because the “glu t” experienced in the Japanese newcar market in 1981 conceivably could have been more se vere without the VERs, these simula tions may under state the VERs ’ influence. However, observation suggests that such severe surpluses are rare; the possibility of a more dramatic sur plus in 1981 is remote. It is possible that the glut was in part VER-induced, if manufacturers attempted to build inventories in antic ipation of a shortage that never fully materialized. 32 lowered sales by only 4,000 units during the first year, a negligible amount for a market in which sales averaged approximately 1.8 mil lion units in the previous two years. During the second year of the VER pro gram, transactions prices increased $273 as dealers experienced more sizable inventory shortages. Most of this increase reflected dealers’ markups, as wholesale prices of new Japanese cars increased $51. Unit sales fell 78.000 during the second year of the VER program. With the U.S. economic recovery under way in 1983, the VER impact on prices intensified; transactions prices rose $1,114. Again, most of the options-adjusted price in creases reflect dealers’ markups ($956), com pared with an options-adjusted wholesale price increase of $158. As a result, unit sales fell 299.000 units between 1983:IIQ and 1984:IQ. The total, three-year loss in consumers’ surplus resulting from the VER-induced increase in Japanese new-car prices was approximately $2.7 billion (see table 2). Most of the loss was experienced in 1983, when the program was most binding on the U.S. market. Of this total amount, $2.6 billion represents a transfer of purchasing power to producers and dealers of Japanese cars from U.S. consumers who continue to buy Japa nese cars at artificially high prices. Approx imately 80 percent of this income transferred to U.S. dealers of Japanese cars and does not represent a net loss to the United States. Japanese producers received the remaining $401 million. Although most funds eventually will return to the United States via world trade, such a return would be a long-term phe nomenon. Because of the VERs, the prices of U.S. imports could rise relative to the prices of U.S. exports, producing a loss to the United States. Of the total reduction in consumers’ surplus, we attribute $177 million to increased inefficiencies in production and foregone consumption opportunities, also representing a loss to the United States. We also measured the net change in rev enues accruing to the Japanese automobile Federal Reserve Bank of Cleveland producers and U.S. dealers because of the VERs. Over the three-year period, Japanese producers and U.S. dealers lost $125 million in revenues because of the VER program; higher prices on the units sold did not gen erate enough revenue to offset a reduction in volume, because the import demand curve is elastic. Without knowledge of Japanese cost curves, however, we do not know how these revenue reductions translate into profit performance. Impact on the U.S. Market Using the estimates obtained from the Japa nese auto demand model, we can speculate about the largest possible effects of the VER program on the number of U.S. cars produced and the number of U.S. workers hired. A fairly standard assumption in the literature on automobile demand is that the overall price elasticity of demand for new cars equals 1. This implies that the total amount of rev enue spent on new cars does not change over all. As an extreme case, we assume that any revenue not spent on new Japanese cars is spent on other new cars. We further assume that revenue not spent on new Japanese cars is spent entirely on new U.S. subcompact cars. This is an overestimate, as consumers would spend some of this revenue on new non-Japanese foreign cars and on compact or full-sized American cars. Dividing this reve nue transfer by the average price of a new subcompact American car, we obtain a rough estimate of the additional units of new U.S. cars produced. Following this procedure, we determined that the VER program increased U.S. car production by 399 units in 1981, 3,444 units in 1982, and 16,768 units in 1983 (see table 3). Having estimated the units produced, we can determine the associated employment effects using assumptions about labor pro ductivity and average hours worked. Table 3 shows the total hours necessary to produce the additional units, adopting a U.S. Con gress (1982) estimate that it takes 200 man- 22. The man-hours estimate refers to autoworkers and workers in directly related industries (see U.S. Congress 1982). hours to produce one subcompact car in the are based on our model’s results, they too United States.22 We estimate that the VER are small relative to the findings of the other program stimulated an additional 79,800 studies cited. production man-hours during its first year, These employment gains do not represent 688,800 hours in 1982, and 3.4 million hours net benefits to the United States because of in 1983. The link between our hours esti the VER program, especially in the long run. mates and actual employment gains is uncer As discussed earlier, the U.S. revenue gains tain, as additional hours may be distributed represent a transfer from consumers to domes among the existing workforce. As a conser tic producers and workers. These funds now vative estimate, we can assume that the remain in the United States and increase average hours worked per U.S. autoworker jobs in the automobile industry, but this does remain unchanged as a result of the export not necessarily imply a net increase in U.S. restraints, in which case additional U.S. employment in the long run. Most funds sent employment can be determined using the abroad to pay for Japanese imports eventually average hours worked per U.S. autoworker. return to the United States as foreigners buy We estimate that VERs increased U.S. auto U.S. exports. In the long run, any gains in mobile employment by 38 workers in 1981, auto industry employment because of the 328 in 1982, and 1,492 in 1983. This compares VERs must be compared with possible losses with indefinite layoffs of 250,000 U.S. auto in U.S. employment among export-oriented workers at the industry’s 1982 employment industries. The net result depends on the trough. Because these employment estimates decline in exports and the relative intensity of labor in the production functions of these two industries. Such comparisons are beyond the scope of this article, but gains in auto Table 1 Impacts of VERs on Prices worker employment should not be labeled and Quantities of New Japanese Cars as net benefits in the absence of such 1981 1982 1983 comparisons. Dealer cost, in dollars With VERs W ithout VERs Difference Percent change Transactions price, in dollars With VERs W ithout VERs Difference Percent change Unit sales estim ates, in thousands With VERs W ithout VERs Difference Percent change 5,479 5,469 10 0.2 5,759 5,709 51 1.1 6,294 6,137 158 2.6 6,284 6,273 0.2 6,941 6,668 273 4.1 8,282 7,168 1,114 15.5 1,962 1,966 -4 -0.2 1,829 1,908 -78 -4.1 1,831 2,130 -299 -14.0 11 NOTE: Annual data are derived from quarterly estimates and may not sum exactly because of rounding errors. Years correspond to VER years, beginning in the second quarter of the current year and running through the first quarter of the subsequent year. 33 Economic Review • Summer 1984 Results in Perspective To the best of our knowledge, ours is the only study that considers the effects of the VER program over its first three years. More over, it is difficult to compare our results with those of other studies, because of meth odological differences (see Crandall 1984; Feenstra 1982; Gomez-Ibanez, Leone, and O’Connell 1983; Stuchlak, Shickman, and Pochiluk, Jr. 1983; and Wharton 1983). Our results show substantially weaker first-year effects than other studies but a large impact by 1983. Two factors seem to distinguish our results from those of earlier studies. Un like other researchers (except Feenstra), we adjusted wholesale prices for options. As dis cussed earlier, studies that fail to adjust for options will overstate the impact of the VER program on prices and measures derived from prices. We also introduced the role of inven tories into the analysis. This should dampen the estimated impacts of VERs, especially Feenstra and Gomez-Ibanez, Leone, and in 1981 and 1982, relative to the results of O’Connell provide sufficient description studies that implicitly assume no inventories. to make some useful comparisons with our results. After adjusting for inflation, Feen stra observes that the average import price of new Japanese cars rose 8.4 percent in 1981. Table 2 Income Transfers and He provides a detailed explanation of quality Efficiency Losses Associated with VERs upgrading and concludes that 5.3 percent In millions of dollars of the observed price increase reflects VER1982 1983 Total 1981 induced quality adjustments. The remaining Total consum ers’ 21.6 510.8 2,206.4 2,738.8 3.1 percent increase is the VER-induced rents surplus loss component of the price rise. Assuming ownIncome transfers price elasticities for Japanese automobile ser 19.4 92.6 288.8 400.8 To Japanese vices of 2 and 3, Feenstra estimates that 1981 m anufacturers sales of Japanese cars fell 220,000 units and 2.2 407.5 1,751.2 2,160.9 To U.S. dealers 277,000 units, respectively. The resulting 500.1 2,561.7 21.6 2,040.0 Total gains in U.S. autoworker employment were 10.7 166.4 177.1 0.0 Efficiency loss 5,600 and 11,100 workers, respectively. After 21.3 101.7 125.3 2.3 M arket revenue adjusting for the influence of quality, Feen loss stra calculates that the consumers’ surplus NOTE: Annual data are derived from quarterly estimates and may not loss was between $322 million and $327 mil sum exactly because of rounding errors. Years correspond to VER years, lion in 1981 under his respective elasticity beginning in the second quarter of the current year and running through assumptions. Feenstra also examines the the first quarter of the subsequent year. effects of VERs under the assumption of an inelastic own-price demand for Japanese automobile services (0.9). In this case, sales Table 3 Impacts of VERs on of Japanese cars would have fallen (123,000 U.S. Autoworker Employment units); total revenues spent on Japanese cars Output, hours, or would have risen. Consequently, U.S. new1983 employment 1981 1982 car sales would have declined 5,300 units and Revenues lost by 2,300,000 21,300,000 101,700,000 U.S. autoworker employment would have Japan, dollars fallen by 600 workers. The total loss in con 6,065 U.S. average 5,768 6,184 sumers’ surplus for the inelastic scenario subcompact equaled $314 million. price, dollars Gomez-Ibanez, Leone, and O’Connell con 3,444 16,768 Additional U.S. 399 structed a model of the U.S. market to mea supcompact production, units sure the effects of the VER program that did not include a quality-adjustment allowance or 688,800 3,353,600 Additional U.S. 79,800 an inventory influence. Instead, they divided production hours the U.S. market into basic small cars (Japa 2,247.7 2,103.8 Average annual 2,125.9 nese and all others), luxury small cars (Japa man-hours per worker nese and all others), and traditional cars. The researchers simulated their model, 1,492 Additional U.S. 328 38 auto employment which is not specific to a particular year, under alternative assumptions about the over NOTE: Annual data are derived from quarterly estimates and may not sum exactly because of rounding errors. Years correspond to VER years, all strength of the U.S. new-car market and beginning in the second quarter of the current year and running through different price/quantity reactions to the VER the first quarter of the subsequent year. program from domestic car producers. The case of a weak U.S. market (total new-car sales of 8.8 million units) and both price and quantity responses from U.S. manufacturers Federal Reserve Bank of Cleveland 23. Actual new-car sales were 8.1 m il lion units in 1981, 8.0 million units in 1982, and 9.2 mil lion units in 1983. seems the most representative of the actual market environment.23 In this case, the VERs raised Japanese new-car prices 2.6 percent per year and reduced Japanese new-car sales in the United States 6.7 percent per year. U.S. car production rose 0.5 percent, and U.S. auto worker employment increased 6,500 workers. Gomez-Ibanez, Leone, and O’Connell esti mated that the loss to consumers in all seg ments of the market was $566 million per year. IV. Conclusion International trade theory demonstrates that artificial barriers against imports raise prices of traded goods, transfer income from con sumers to producers, and create production and consumption inefficiencies. This article has illustrated these effects for the case of the Jap anese VERs on new-car imports to the United States. The results of our empirical analysis are tentative approximations because of the small size of our sample and the unavoidable difficulties of estimating structural models. Given the partial equilibrium nature of our analysis, the costs of the VER program could be greater than we have estimated. Our results suggest that in its initial year the VER program had little effect on the U.S. market for Japanese cars and did not appreciably create new auto-industry employ ment in the United States. At the time, inven tories of new Japanese cars were overstocked because of weakening new-car demand and high-inventory carrying costs. With the U.S. economic recovery under way in 1983, inven tory shortages at the dealers’ level became extreme. According to our partial equilibrium estimates, the VER program so far has cost the United States approximately $2.7 billion in lost consumers’ surplus. The VER program is an expensive way to increase U.S. auto industry employment. Such expensive walls for the U.S. auto industry surely must give offense to U.S. consumers. Economic Review • Summer 1984 Appendix 5= Japanese new-car sales are measured nonseasonally adjusted in thousands of units (data from Ward’s Automotive Reports). The sales data include captives (Japanese new cars sold by U.S. manufacturers), since the captive market is included in the VERs. Regardless, the captive share of the Japanese new-car mar ket historically has been quite small (less than 5 percent). Per capita auto sales, SALES, were found by dividing unit sales by the total U.S. civilian, noninstitutional population. sales forecast variable, Sf, is derived from a specification similar to that found in Irvine (1979): Set = SM + Sm (Sm ~ S,-s)/Sm>. /= inventories are nonseasonally adjusted, measured in thousands of units, from Ward’s Automotive Reports (quarterly data, 1974 to 1984). VJ72 = nonseasonally adjusted wholesale price index of Japanese cars was obtained from the Bureau of the Census, U.S. General Imports: Sched ule A, commodity by country (FT135). The price was determined by dividing the total customs value (c.i.f.) of new Japanese cars and utility vehicles, including shipping costs to the port of entry (c.i.f.) by the volume of Japanese new-car imports. The average, nom inal wholesale price was deflated, using the implicit price deflator for personal consump tion expenditures. Y 72 = the income variable used in this analysis was real permanent income, calculated as the fit ted quarterly values of real disposable income regressed against four years of lagged values. The income variable was generated in per capita form, using the total U.S. noninstitu tional civilian population. C 72 = real marginal costs of new Japanese car inventories was calculated as the product of real dealer new-car costs (VJ72) and the nonseasonally adjusted 90-day commercial paper rate, i, or VJ72 • i. OPTION = the options index was calculated with the following equation: 5 Index = ^ i=1 Wi ■ Oi, and domestic subcompacts would be useful. Data that would allow such comparisons are not readily available. Indeed, it is unclear that significant operating cost differences between the two markets exist. YEN = the dollar/yen exchange rate is nonseasonally adjusted. References Bergsten, C. Fred. On the Non-Equivalence of Import Quotas and “ Voluntary” Export Restraints. Technical Series Reprint T-009. Washington: Brookings Institution, 1975. where the weights, Wiywere determined according to each options 1980 list price relative to the Carlson, Rodney L. “Seemingly Unrelated total options package costs: Regression and the Demand for Automo biles of Different Sizes, 1965-75: A Disag w, gregate Approach,” Journal of Business, Automatic transmission 0.174 vol. 51, no. 2 (April 1978), pp. 243-62. Power steering 0.087 Carlson, Rodney L., and M. Michael Umble. Air conditioning 0.283 “Forecasting the Demand for Automobiles, Sunroof 0.348 1983-1985: A Disaggregate Approach,” Audio system 0.108 Akron Business and Economic Review, 1.000 vol. 14, no. 4 (Winter 1983), pp. 35-41. The options data, O,, represent the percen tage of each option installed in Japan relative Chow, G.C. Demand for Automobiles in the United States: A Study in Consumer to total Japanese imports. Data on options Durables. Amsterdam: North Holland installations are available semiannually over Publishing Company, 1957. the 1975 to 1983 period. The index was then linearly interpolated to create a quarterly Crandall, Robert W. “Import Quotas and the time series. Automobile Industry: The Costs of Pro tectionism.” Unpublished manuscript, to GAS7 2 = be published in Brookings Review, Brook for an approximate measure of Japanese car ings Institution (Summer 1984). operating costs, we generated an index of aver age car operating expenses relative to the Currie, John Martin, John A. Murphy, and total consumer price index (CPI). The compo Andrew Schmitz. “The Concept of Eco nents of the CPI include gasoline (weight nomic Surplus and Its Use in Economic = 0.66), auto insurance (weight = 0.18), and Analysis,” Economic Journal, vol. 81, auto repairs (weight = 0.16). Each weight no. 324 (December 1971), pp. 741-91. was determined by the component’s relative importance in the CPI (December 1982). Ideally, a variable that also captures oper ating efficiency differences between Japanese Federal Reserve Bank of Cleveland Fair, Ray C. “The Estimation of Simultane ous Equation Models with Lagged Endog enous Variables and First-Order Serially Correlated Errors,” Econometrica, vol. 38, no. 3 (May 1970), pp. 507-16. Falvey, Rodney E. “The Composition of Trade within Import-Restricted Product Cate gories” Journal of Political Economy, vol. 87, no. 5 (October 1979), pp. 1105-14. Feenstra, Robert C. “Voluntary Export Re straints in U.S. Autos, 1980-1981: Quality, Employment and Welfare Effects.” Inter national Economics Research Center Paper 17, 1982. Feldstein, Martin, and Alan Auerbach. “Inventory Behavior in Durable-Goods Manufacturing: The Target-Adjustment Model,” Brookings Papers on Economic Activity, 2:1976. Gomez-Ibanez, Jose A., Robert A. Leone, Stephen A. O’Connell, “Restraining Auto Imports: Does Anyone Win?” Journal of Policy Analysis and Management, vol. 2, no. 2 (Winter 1983), pp. 196-219. Griliches, Zvi, Ed. “Hedonic Price Indexes for Automobiles: An Econometric Analysis of Quality Change,” in Price Indexes and Quality Change. Cambridge: Harvard University Press, 1971. Irvine, F. Owen, Jr. A Study of Automobile Inventory Investment. Working Paper No. 6. Washington: Board of Governors of the Federal Reserve System, December 1979. Loos, Susan A. “The Japanese Cost Advantage in Automobile Production,” Economic Commentary, Federal Reserve Bank of Cleveland, July 2, 1984. Economic Review • Summer 1984 Morkre, Morris E., and David G. Tarr. The Effects of Restrictions on United States Imports: Five Case Studies and Theory. Staff Report of the Bureau of Economics to the Federal Trade Commission. Wash ington: U.S. Government Printing Office, June 1980. Rodriguez, Carlos Alfredo. “The Quality of Imports and the Differential Welfare Effects of Tariffs, Quotas, and Quality Controls as Protective Devices,” Cana dian Journal of Economics, vol. 12, no. 3 (August 1979), pp. 439-49. Santoni, Gary J., and T. Norman Van Cott. “Import Quotas: The Quality Adjustment Problem,” Southern Economic Journal, vol. 46, no. 4 (April 1980), pp. 1206-11. Smith, R.P Consumer Demand for Cars in the USA. Cambridge: Cambridge University Press, 1975. Stuchlak, Wesley J., Daniel E. Shickman, and William R. Pochiluk, Jr., “The Impact of Japanese Import Restraints,” Chase Econ ometrics: Automotive, First-quarter 1983. Tishler, Asher. “The Demand for Cars and the Price of Gasoline: The User Cost Approach.” Review of Economics and Sta tistics, vol. 64, no. 2 (May 1982), pp. 184-90. U.S. Congress. U.S. House of Representa tives. Subcommittee on Trade of the Committee on Ways and Means. Domes tic Content Legislation and the U.S. Auto mobile Industry: Analyses of H.R. 5133. August 1982. Wharton Motor Vehicle Service. “Special Anal ysis: The Japanese Quota,” January 1983, pp. 31-33. Willig, Robert D. “Consumer’s Surplus With out Apology,” American Economic Review, vol. 66, no. 4 (September 1976), pp. 589-97. Owen F. Humpage is an economist with the Federal Reserve Bank of Cleveland. Working Paper Review Owen F. Humpage Dollar Intervention and the Deutschemark-Dollar Exchange Rate: A Daily Time Series Model Working Paper 8404. September 1984. 28 pp. Bibliography. In March 1973, the major industrialized nations abandoned the Bretton Woods fixedexchange-rate system. Observers have charac terized the subsequent exchange-rate regime as a dirty float. While the major industrialized countries generally have allowed fundamen tal market forces to determine their exchange rate, they periodically have bought and sold foreign exchange to influence the market out come. The volume and frequency of exchangemarket intervention have varied greatly among the developed countries. Economists have questioned the efficacy of foreign-exchange-market intervention, espe cially if intervention is divorced from monetary policy (sterilized) and especially if exchange markets are highly efficient. An emerging consensus holds that sterilized intervention has no lasting impact in foreign-exchange markets and cannot be used to supplant the impact of such market fundamentals as rel ative money-stock-growth rates, inflation rates, or interest rates. Researchers, however, have not rejected the possibility that sterilized inter vention has a short-run or temporary impact on exchange rates. In this working paper we develop a simul taneous time-series model to investigate the daily interactions between U.S. exchangemarket intervention and the deutschemarkdollar exchange rate. Such an investigation involves answering two questions. How does the Federal Reserve System react to exchangerate movements? How does the exchange 38 Federal Reserve Bank of Cleveland rate respond to intervention? By incorporat ing both a morning-opening and an after noon-closing deutschemark-dollar quote, and by assuming that U.S. intervention occurs only during the interim, this study attempts to interpret the direction of causality be tween contemporary exchange-rate move ments and intervention. The model divides U.S. intervention into dollar purchases or sales of deutschemark and dollar purchases or sales of all other currencies to capture both the direct and cross-rate effects of interven tion. The model also includes an aggregate measure for the intervention of the other large developed countries. Using autoregres sive integrated moving average (ARIMA) tech niques, we estimate the model from Novem ber 2, 1978, to October 31, 1979, a period of frequent, often heavy intervention. The model indicates that, on average over the period investigated, U.S. intervention re acted without a lag to unanticipated changes in the morning-opening exchange rate in a manner consistent with a leaning-againstthe-wind strategy. Such a strategy would tend to dampen exchange-rate fluctuations if it were capable of altering the exchange rate in the appropriate direction. The results, however, do not indicate that intervention, on average over this period, was effective in changing the exchange rate in the desired direction. The coefficients on the intervention terms suggest that U.S. and foreign interven tion accentuated movements in the afternoonclosing exchange rate. The size of the effect, however, was very small. The seemingly perverse exchange-rate response could be rational from the perspec tive of private exchange-market participants. Foreign-exchange speculators could view central-bank support for a currency as under scoring that currency’s fundamental weak ness, and they could react by selling that cur rency. Moreover, the experiment’s failure to find the expected exchange-rate response to intervention could result because the volume of intervention was too small relative to the volume of daily exchange-market transactions or because the influence of daily intervention deteriorates too quickly to be detected by the closing quote. The Federal Reserve Bank of Cleveland publishes an infor mative research peri odical called Eco nomic Commentary. Following are the titles published since January 1984. If you are interested in receiving this publi cation, either future or back issues, please contact our Public Information Center, Federal Reserve Bank of Cleveland, P.O. Box6387, Cleve land, OH 44101, (216) 579-2048. Economic Commentary The Economy in 1984: Industry Perspectives Robert H. Schnorbus 5/7/84 The International Debt Situation Owen E Humpage 1/3/84 Nominal Income Targeting John B. Carlson 5/21/84 Commercial Bank Holdings of Treasury Debt Gary Whalen 1/16/84 Seeking a Stable Economic Environment Karen N. Horn 6/4/84 Closely Watched Banks Paul R. Watro 1/30/84 Regional Interstate Banking Gerald H. Anderson, Thomas M. Buynak, and James J. Balazsy, Jr. 6/18/84 Collective Bargaining and Disinflation Mark S. Sniderman and Daniel A. Littman 2/13/84 The Japanese Cost Advantage in Automobile Production Susan A. Loos 7/2/84 Monetary Policy in the 1980s Karen N. Horn 2/27/84 Rate Deregulation and Deposit Shifting Paul R. Watro 7/16/84 Banking without Interstate Barriers Thomas M. Buynak, Gerald H. Anderson, and James J. Balazsy, Jr. 3/12/84 The Costs of a Protectionist Cure Michael F. Bryan and Owen F. Humpage 7/30/84 The Monetary Targets in 1984 William T. Gavin 3/26/84 Small-Issue IDBs—Tax Policy in Search of a Focus Paul Gary Wyckoff 8/13/84 Reorganizing the U.S. Banking Regulatory Structure Sandra Pianalto 4/9/84 Deregulation and Deposit Pricing Paul R. Watro 4/23/84 Economic Review • Summer 1984 The Recovery of Durable Goods: What Exhilarated the Consumer? Lawrence Slifman 8/27/84 Economic Review is published quar terly by the Research Department of the Federal Reserve Bank of Cleveland. Copies of the issues listed here are available through our Public Information Center, 216/579-2048. Economic Review Winter 1983 Location and Reinvestment: The Youngstown Steel District Robert H. Schnorbus Thrifts and the Competitive Analysis of Bank Mergers Paul R. Watro Working Paper Review: Stability in a Model of StaggeredReserve Accounting Michael L. Bagshaw and William T. Gavin Spring 1983 Money Demand: Cash Management and Deregulation John B. Carlson Divisia Monetary Aggregates: Would They Be More Palatable Than the Traditional Simple-Sum Stews? Mark A. Zupan Fall 1983 Plant Closings and Worker Dislocation Daniel A. Littman and Myung-Hoon Lee Prevailing Wage Laws, the Federal Reserve, and the Service Contract Act Mark S. Sniderman Working Paper Review: Forecasting the Money Supply in Time Series Models Michael L. Bagshaw and William T. Gavin W inter 1984 Reflections on Money and Inflation William T. Gavin The Outlook for Inflation K.J. Kowalewski and Michael F. Bryan Spring 1984 Estimating Infrastructure Needs: Methods and Controversies Paul Gary Wyckoff Nonbanking Operations of Bank Holding Companies Gary Whalen 40 Federal Reserve Bank of Cleveland