The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
3 T h e Internationalization o f the Beer B rew in g Industry 20 Alternative Measures o f M on ey as Indicators o f Inflation: A Survey and Some N e w Evidence 34 T h e Trade-Related Aspects o f Intel lectual P ro p e rty Rights: W h at Is At Stake? 47 T h e Economic Consequences o f Reducing M ilitary Spending THE FEDERAL A RFSEim J&HXSWoi A r S T .m r is 1 F ederal R eserve Bank o f St. Louis R eview November/December 1990 In This Issue . . . The brew in g industry has evolved over the past 25 years from an in dustry that concentrated on domestic markets to one that view s itself as part o f a global market. In the first article o f this Review, Jeffrey D. Karrenbrock examines "Th e Internationalization o f the Beer Brewing In dustry.” Th e author provides a case study o f h ow and w hy brew ers have used merchandise trade, licensing agreements and foreign direct investment to penetrate foreign markets. In addition, Karrenbrock demonstrates how certain economic factors, such as economies o f scale and trade barriers, can affect which type o f international transaction is used in different markets. A fe w specific countries’ barriers to beer trade are also discussed. * * * On October 6, 1979, the Federal Reserve announced its landmark deci sion to implement m onetary policy by targeting the grow th rate o f M l rather than some level o f the federal funds rate. Th e Federal Reserve abandoned this practice less than three years later prim arily because empirical regularities in the behavior o f M l velocity that had made M l targeting an attractive strategy seemed to disappear. Although no one has yet offered a definitive explanation fo r the observed changes in M l velocity, these changes nonetheless have led to a general discrediting o f m onetary aggregate targeting as an approach to m onetary policy. In the second article o f this issue, "Alternative Measures o f M oney As Indicators o f Inflation: A Survey and Some N ew Evidence,” Michael T. Belongia and James A. Chalfant review the arguments about w h y M l velocity may have changed, then discuss m ore general issues about the construction o f a measure o f the money stock and the properties one w ould expect it to exhibit. Noting that the composition and weighting o f components in a m onetary aggregate each may contribute to measure ment erro r and unexplained shifts in behavioral relationships, the authors then review other recent efforts to construct new m onetary ag gregates from first principles o f economic and statistical theory. They compare the empirical perform ance o f the alternatives against various final use criteria and find that, although some measures o ffe r im provem ents over the current official aggregates, none can be judged at this time to be the definitive measure o f the m oney stock. * * * In the third article in this issue, "Th e Trade-Related Aspects o f In tellectual Property Rights: W hat Is At Stake?” Alison Butler examines one o f the major n ew issues being negotiated in the current round o f multilateral negotiations. W ith increased trade in high-technology pro ducts, intellectual property rights have becom e an increasingly im por tant and contentious issue in international trade. The protection o f in- NOVEMBER/DECEMBER 1990 FEDERAL RESERVE 2 tellectual property, w hich includes such things as patents, copyrights and trademarks, is justified by most economists as being necessary fo r invention, w hile others, prim arily policymakers, argue that these rights only serve to protect the m onopoly pow er o f innovating firms. The author analyzes the questions surrounding the protection o f in tellectual property, how innovation is affected by trade, and the benefits o f protecting intellectual property internationally. In addition, she ex amines w h y the incentive to protect intellectual property differs betw een industrialized and developing countries and what likely needs to happen b efore an agreem ent can be reached. * * * Before the crisis in the Middle East this summer, the easing o f interna tional tensions had reduced, fo r many, the urgency fo r the United States to continue building or even maintain its m ilitary strength. As support fo r allocating the nation’s resources to defense weakened, peo ple began to argue about the potential fo r a significant “ peace dividend” available to the U.S. economy. In the fourth article o f this issue, "The Economic Consequences o f Reducing M ilitary Spending,” Michelle R. Garfinkel examines some issues that have been overlooked by those searching fo r ways to use the peace dividend. Garfinkel attempts to answer tw o important economic ques tions: How can the resources from reduced m ilitary spending be real located efficiently to nonmilitary uses, and h ow can w e identify what these uses should be? W hile the recent course o f events in the Middle East might appear to detract from the prospects o f a large dividend in the near future, it does not detract, the author says, from the relevance o f these questions. Assuming that conflict is not a permanent obstacle to realizing a large dividend later, it provides us with m ore time now to evaluate the various options associated w ith future defense cuts. * * * BANK OF ST. LOUIS 3 Jeffrey D. Karrenbrock Jeffrey D. Karrenbrock is an economist at the Federal Reserve Bank of St. Louis. David H. Kelly provided research assistance. The Internationalization of the Beer Brew ing Industry T HE BEER BREWING INDUSTRY has been undergoing a process o f internationalization fo r the past 25 years. This article examines the roles that three types o f international transactions — merchandise trade, licensing agreements and foreign direct investment — have played in this internationalization. As in other industries, a fe w general economic factors explain much o f the increase in international brew in g activity. W hat makes beer brew in g a particularly inter esting case study is that it provides an oppor tunity to demonstrate h ow certain economic factors, such as economies o f scale and trade barriers, can affect the internationalization o f an industry. THE INTERNATIONALIZATION OF THE BREWING INDUSTRY Merchandise Trade As w ith most other goods, w orld merchandise trade in b eer has expanded rapidly over the past 25 years (see figure 1). Much o f the in crease in w orld beer trade — and in w orld trade in general — can be attributed to such factors as low er trade barriers, m ore efficient communication and transportation technology, and grow th in real personal incomes. Th e value o f w orld beer trade increased from $149 million in 1965 to $2.08 billion in 1987, a 14-fold in crease; at the same time, w orld trade in all goods increased to m ore than 12.5 times its 1965 value. In m ore recent years, betw een 1980 and 1987, w orld trade in beer expanded 83.8 percent, w hile total w orld trade g rew only 23.4 percent. Despite its rapid growth, trade in beer in 1987 accounted fo r less than one-tenth o f one percent o f total w orld merchandise trade. On a volume basis, w orld trade in beer has nearly tripled since 1965, grow in g at an average annual rate o f 6.5 percent betw een 1965 and 1987. The largest exporters o f beer in this grow in g market, ranked by volume, are the Netherlands, W est Germany, Czechoslovakia, Belgium and Canada (see table l ) . 1 The largest im porters are the United States, the United Kingdom, France, Italy and W est Germany. Beer imports as a percent o f total consumption (IPC) and exports as a percent o f production (EPP) are larger fo r some o f the smaller exporters and im porters than they are fo r some o f the larger exporting and im porting countries. As table 1 shows, among 25 importers, beer IPC ranges from a low o f 0.2 percent in N orw ay to 16.4 percent in Italy. The percent o f beer con sumption accounted fo r by imports in the largest beer im porting country, the United States, is about 5 percent. Similarly, among exporters, figures fo r EPP range from 0.4 percent in the United States to 'The FAO Trade Yearbook indicates that Mexico was the world’s third-largest exporter of beer in 1987. NOVEMBER/DECEMBER 1990 4 Figure 1 World Merchandise Trade Billions of dollars SOURCES: Millions of dollars FAO Trade Yearbook and International Financial Statistics Yearbook. 41.6 percent in Ireland. The export numbers as a percent o f production fo r such countries as the Netherlands and Luxembourg, how ever, are questionable as these countries do a significant amount o f re-exporting to other countries (that is, much o f the beer reported as exports may simply be im ported and then re-exported fo r consumption elsew here).2 Few o f the countries listed in table 1 are strictly im porters or exporters o f beer. Most o f the countries that export beer also im port some beer and vice-versa. This pattern o f trade is known as intra-industry trade. An examination o f the IPC and EPP statistics in table 1 show 2Ott (1988) notes that the Netherlands has a long history of re-exporting imported goods. 3Ott (1988) notes that re-exported goods from the Netherlands are not included in the country’s import figures. FEDERAL RESERVE BANK OF ST. LOUIS that intra-industry trade in beer is m ore im por tant to some countries than others. The largest exporter o f beer, the Netherlands, im ported on ly 4.3 percent o f the beer it consumed in 1987.3 Similarly, the tw o largest im porters o f beer, the United States and the United Kingdom, exported only 0.4 percent and 1.9 percent o f their beer production in 1987. Ireland, on the other hand, exported nearly 42 percent o f its production, w hile im porting m ore than 12 percent o f its beer consumption. The degree o f intra-industry trade fo r a coun try can be measured using a simple index, cal culated fo r a given product as the absolute value Selected 1987 Brewing Industry Statistics__________________________________________________________ Average Annual Growth Rate, 1975 to 1987 Australia Austria Belgium Canada Czechoslovakia Denmark W. Germany Finland France E. Germany Hungary Ireland Italy Japan Luxembourg Netherlands New Zealand Norway Poland Portugal Spain Sweden Switzerland USSR United Kingdom USA Exports 1000 HL Imports 1000 HL 728.3 361.0 2537.0 2415.8 2698.0 1934.0 5706.0 22.8 672.0 N/A 0.0 2066.0 73.1 293.3 270.3 5725.2 83.2 19.2 327.1 74.5 121.0 44.9 36.7 N/A 1140.0 919.7 70.0 285.0 565.5 448.0 0.0 20.8 1301.6 13.9 2445.9 N/A 1290.0 415.0 2162.8 224.0 42.2 534.9 70.5 405.0 106.2 34.3 735.0 344.4 485.9 900.0 4093.1 10991.1 Exports Imports As Percent of As Percent of Production Consumption 3.9% 4.0 18.1 10.5 12.1 22.8 6.2 0.7 3.4 N/A 0.0 41.6 0.7 0.5 40.9 32.6 2.0 0.9 2.8 1.5 0.5 1.1 0.9 N/A 1.9 0.4 0.4% 3.2 4.7 2.1 0.0 0.3 1.5 0.4 11.3 N/A 12.6 12.5 16.4 0.4 9.8 4.3 1.7 0.2 0.9 0.7 2.8 7.9 10.5 1.7 6.5 5.0 Intra-Industry Trade Index 1975 1987 .83 .33 .23 .63 1.00 .98 .54 .60 .56 .25 N/A .98 .95 .73 .79 .69 .68 .89 .74 .97 .58 .91 .75 N/A .54 .80 .82 .12 .64 .69 1.00 .98 .63 .24 .57 N/A 1.00 .67 .93 .13 .73 .83 .08 .62 .51 .37 .72 .77 .86 N/A .56 .85 Exports Imports Per As Percent of As Percent of Capita Production Consumption Consumption 15.0% 7.4 2.4 13.8 5.9 -0.3 8.2 53.7 2.4 N/A 0.0 2.1 20.5 1.8 -0.7 6.2 7.2 -12.2 2.7 10.9 14.5 72.0 2.8 0.0 3.1 22.1 17.0% -0.7 -5.3 52.0 0.0 * 6.3 171.1 2.4 N/A -2.86** 39.0 5.7 19.6 0.5 2.4 36.7 22.6 12.3 N/A 19.3 8.4 6.4 8.8 3.4 13.7 - 1.7% 1.0 -0 .5 -0 .4 -0 .7 -0 .2 -0 .2 1.9 - 1.1 1.6 2.5 -2 .0 6.9 2.0 -0 .8 0.6 -0 .7 1.1 -1 .5 3.7 2.7 -1 .2 -0 .3 - 1.4 -0 .4 0.8 Imports Per Capita 14.7% 0.2 -5 .7 50.7 0.0 * 6.1 184.6 1.3 N/A -0.83* 35.9 12.9 22.6 -0 .3 2.8 33.9 26.6 10.7 N/A 22.4 7.0 6.1 7.0 2.9 14.7 SOURCE: Derived from information in the Brewers Association of Canada’s International Survey: Alcoholic Beverage Taxation and Control Policies. 1 HL = 100 liters = 26.4 gallons * Denmark’s beer imports were generally small and declining throughout the period. However, a few relatively large increases in imports in later years caused imports as a percent of consumption and imports per capita to grow at average annual rates of 2,218 and 2,146.4, respectively. ** Average annual rate of growth for 1978-1986. NOVEMBER/DECEMBER 1990 Table 1 6 o f the difference betw een exports and imports divided by the sum o f exports and imports.4 If the index is close to zero, the degree o f intra industry trade is substantial. An index value o f one indicates that there is no intra-industry trade. This index, labeled the "Intra-Industry Trade Index” and shown in table 1, was calcu lated fo r beer trade in 1975 and 1987. O f the 23 countries fo r which the index could be calculated fo r both years, nine countries’ in dexes rose over the period, indicating less intra industry beer trade. In 12 countries, the indexes declined, indicating that intra-industry beer trade had increased. Only five countries in table 1 had an intra-industry index value o f less than 0.5 in 1987. The majority (70 percent) had an index value o f m ore than 0.5 in 1987, which indicates that intra-industry trade plays a m inor role in the brew in g industry in general, although it has becom e m ore prevalent during the past 15 years. Th e grow th rates o f IPC and EPP provide fu r ther evidence o f the increasing importance o f intra-industry trade to the brew in g industry. O f the 20 countries in table 1 reporting increased EPP betw een 1975 and 1987, 16 also reported increased IPC. Similarly, o f the 19 countries reporting increased IPC, 16 reported increased EPP. In sum, merchandise trade in beer has ex panded rapidly during the past 25 years, with intra-industry trade playing a small, but growing, role in beer trade. Licensing Agreements Brewers also use licensing agreements to make their products available to foreign consumers. A typical license agreem ent allows a b re w e r in one country to b rew and market the beer o f a foreign brew er. One example is AnheuserBusch’s (A-B) licensing agreem ent w ith the Cana dian b re w e ry John Labatt Ltd. This agreement allows Labatt to b rew and market some A-B beers, such as Budweiser and Michelob, in Cana da. In return, Labatt pays a royalty fe e to A-B. O f course, the licensing brew ers insist that the consistent quality o f their products be main tained. In essence, the licenser is selling its “See Gray (1987), pp. 243-49. 5The U.S.-Canada Free Trade Agreement eliminates the federal tariffs on beer between these countries, but does not alter the pricing practices of the provincial liquor outlets. See Carter, et al (1989) for a more detailed description of Canadian barriers to beer trade. FEDERAL RESERVE BANK OF ST. LOUIS know-how in brew in g a specific beer, the right to use a trademark and the name recognition it has built fo r that tradem ark in exchange fo r a royalty payment from the licensee. There w ere at least 30 licensing agreements among various brew ers around the w orld in 1987 (see table 2). Several factors that are not mutually exclusive prom ote the use o f licensing agreements. First, some firm s use licensing agreements to circum vent trade barriers. For example, U.S. beers that are b rew ed in Canada under license agree ment are not subject to either the Canadian federal ta riff or the discriminatory mark-ups that other im ported beers face at the provincial governm ent outlets.5 Second, the physical quali ties o f beer prom ote the use o f licensing agree ments. Beer is about 90 percent water, so trans portation costs can be reduced through local production. In addition, beer has a shelf-life o f about three to fou r months, o f which tw o to three weeks could be taken up by overseas shipment. Also, w hen companies enter new markets, they often find it m ore profitable to license existing plants and distribution systems to handle their products rather than build their ow n plants and establish their ow n distribution systems. The im port and export figures discussed p revi ously did not include consumption o f foreignheld brand names that are b rew ed domestically under a licensing agreement. Thus, the degree o f internationalization is understated w hen only merchandise trade is analyzed. Inform ation on the amount o f beer b rew ed under licensing agreements is usually closely held b y the com panies involved, and not much data are publicly available. The Conference Board o f Canada, how ever, has estimated the impact o f licensed b rew ed beer in the Canadian beer market and its findings serve to demonstrate how important licensed production can be. In the Canadian market, three U.S. brewers, Anheuser-Busch, Coors and Miller, had licensing agreements w ith the three largest Canadian b rew ers— John Labatt, Molson and Carling O’K eefe— respectively, in 1986.6 Th e Conference 6Molson and Carling O’Keefe agreed to merge their breweries in 1989. The new company is called Molson Breweries and will continue to brew for Coors and Miller under license. 7 Table 2 Licensing Accords Between Major Brewers Anheuser-Busch Heineken Kirin Guinness Kirin brews Heineken in Japan Guinness brews Bud in Ireland Labatt Whitbread Labatt brews Bud, Michelob & Bud Light in Whitbread brews Heineken in U.K. Canada Kirin National National brews Bud in Israel Molson Molson test brewing Kirin in Canada** Oriental Oriental brews Bud in Korea Lowenbrau Suntory Allied-Lyons Suntory brews Bud in Japan Allied-Lyons brews Lowenbrau in U.K. United Asahi United brews Bud in Denmark Asahi brews Lowenbrau in Japan Watney Mann Miller Watney brews Bud in U.K. Miller brews Lowenbrau in U.S. Artois Molson Whitbread Molson brews Lowenbrau in Canada Whitbread brews Stella in U.K. Miller Carlton & United* Carling O’Keefe' Carling O’Keefe* Carling brews High Life, Lite in Canada Carling brews Foster's in Canada Courage * Courage * Courage brews Lite in U.K. Courage brews Foster’s in U.K. United Watney Mann Watney brews Foster’s in U.K. Beamish & Crawford Beamish & Crawford brews Carlsberg in Castlemaine Parkins Ireland Allied-Lyons Carling O’Keefe * Allied brews XXX in U.K. Carling brews Carlsberg in Canada1 Coors Falken Falken brews Carlsberg in Sweden Asahi Asahi brews Coors in Japan Heileman Molson Heileman brews Tuborg in U.S. Molson brews Coors in Canada Suntory Suntory brews Carlsberg in Japan Guinness Labatt Watney Mann Labatt brews Guinness in Canada Watney brews Carlsberg in U.K. ’ Parent company, Elders IXL. **No licensing agreement in effect. SOURCE: Modem Brewery Age, July 13, 1987. 'John Labatt Ltd. acquired the right to brew Carlsberg in Canada, effective July 1, 1988. NOVEMBER/DECEMBER 1990 8 Board estimates that brands produced in Canada under license w ith U.S. brew ers in 1986 may have accounted fo r as much as 15 percent o f beer sales in Canada.7 This amounts to approx imately 2.7 million barrels o f U.S. brands pro duced and sold under license agreements in Canada in 1986. I f these estimates are correct, the volume o f licensed production o f U.S. beers in Canada was m ore than 17 times the amount o f beer exported directly to Canada in 1986 and m ore than four times the amount o f total U.S. beer exports to all countries (exclusive o f ship ments to U.S. m ilitary bases and Puerto Rico).8 In terms o f Canadian consumption, the licensed b rew ed beer might have accounted fo r 15 per cent o f Canadian beer consumption compared with the 2.1 percent o f domestic consumption accounted fo r by imports. The numerous licensing agreements with brew eries in Japan and the United Kingdom might indicate that beer produced under license represents a significant part o f the foreign beer consumed in these countries. At least in some countries, beer produced under license clearly accounts fo r a much larger portion o f foreign beer consumption than does im ported beer. Foreign Direct Investment In addition to merchandise trade and licensing agreements, the internationalization o f the b re w ing industry has been characterized b y the in creasing production o f beer by "foreign-ow ned” firms. This production reflects the increasing frequency o f foreign direct investment (FDI), in which one b re w e r purchases an existing firm or invests in a new or existing facility in a foreign country. Like licensing agreements, FDI is a substitute fo r merchandise trade. Firms may be prom pted to use FDI fo r the same reasons they use licensing agreements. In addition, such fac tors as low er labor and energy costs and less governm ent regulation may also encourage the use o f FDI. Several b rew ers have invested capital in brew eries outside their home countries. T w o Australian brew ers, Elders IXL and Bond, have Conference Board of Canada (1989), p. 9. 8Data for U.S. beer exports to Canada and total U.S. beer exports, exclusive of shipments to military bases, Puerto Rico and the territories, were provided courtesy of R.S. Weinberg & Associates. 9Carrington (1989). FEDERAL RESERVE BANK OF ST. LOUIS used this method o f globalization extensively. Elders purchased Courage Ltd. o f England in 1986, then purchased Carling O ’K eefe breweries o f Canada in 1987. Early in 1989, Britain's De partment o f Trade and Industry blocked the proposed takeover o f Scottish & Newcastle Breweries PLC by Elders IXL Ltd.9 In 1990, Elders IXL announced that its United Kingdom Courage brew eries w ould purchase Grand M etropolitan’s U.K. brew ing and brands interests and its beer distribution and wholesaling activities. Further more, Courage’s 4,900 pubs w ill m erge w ith Grand M et’s 3,570 pubs in a joint venture under the name Inntrepreneur Estates.10 Bond has purchased tw o U.S. breweries, Pittsburgh B rew ing and G. Heileman. Japanese b rew ers also have actively invested in foreign b re w e ry operations. In 1989, Tatsuuma-Honke Brewing Co. announced plans to build a sake b re w e ry on the grounds o f the Coors b rew ery in Colorado. In 1990, Asahi Breweries Ltd. announced plans to invest $70 million to open a b re w e ry near Denver, Colorado, w h ere it w ill produce a dry b eer.11 Finally, the Canadian brew ing company John Labatt Ltd. purchased Latrobe Brewing Co. o f the United States in 1987. THE ECONOMICS OF INTERNA TIONALIZATION Underlying the preceding description o f the internationalization o f the brew ing industry are some economic factors. The next section outlines the reasons w h y demand fo r foreign beer can exist in a country that already produces some domestic brands and discusses how changing relative prices and rising income can expand the demand fo r foreign beer. The second sec tion analyzes the basic economic factors that determine the type o f international transaction a b rew ery w ill use to put its products in the hands o f foreign consumers. A m ore technical presentation o f the economics discussed in these tw o sections is provided in appendixes to this article. 10See Thornhill and Harris (1990), Harris (1990) and Sherwell (1990). " “ Japan’s Asahi Plans Brewery in U.S.” (1990). 9 Foreign Demand: The Attributes Approach the low-calorie beer the rest o f the time. Why Demand f o r Foreign B eer Can Exist— The demand o f foreign beer can expand if its price falls relative to the price o f domestic beer. I f a consumer had been purchasing domestic beer, the relative fall in the price o f the foreign beer may be enough to compensate him fo r any perceived loss in services due to switching from the domestic to the foreign brand. In this case, the quantity demanded o f the foreign beer w ill increase. Th e decline in the relative price o f the foreign beer may also encourage people w ho already consume it to purchase more. U n for tunately, data on im ported beer prices are scarce, and thus the role that changing relative foreign beer prices has played in the globaliza tion o f the industry is uncertain. One reason w hy people consume foreign beer is that they can buy it at a price at w hich they w ant m ore beer than domestic brew ers want to produce. That is, the quantity o f beer demand ed is larger than the quantity o f beer supplied domestically at the price o f foreign beer, and therefore, some foreign beer is im ported to meet the excess demand. Another reason w hy people consume foreign beer is that at least some consumers p refer the attributes, or char acteristics, o f the foreign beer over domestic brands. This second possibility is discussed in this section. In general, consumers purchase beer fo r the "services” that they feel it can provide.12 Con sumers have a w ide variety o f beer brands to choose from and, subject to price and income limitations, w ill choose those brands that have the attributes that most closely match their de sired services from drinking beer. Many attri butes, such as taste, caloric content, alcohol content and packaging, distinguish one brand from another, and each combination o f charac teristics offers a distinctly different package of services. Brewers differentiate their products on the basis o f attributes and price. Consumers com pare the package o f services provided by a par ticular beer and its price to the services and prices o f other brands. If consumers p refer the services o f foreign beer over domestic beer, at given market prices, then demand fo r foreign beer w ill exist in a country. O f course consumers do not necessarily con sume only one domestic beer or foreign beer. Consumer satisfaction may be maximized by purchasing a combination o f domestic and fo r eign beers. Suppose a consumer prefers the taste o f a high-calorie foreign beer over all other domestic brands, but needs to watch his caloric intake and finds the taste o f a particular brand o f light beer to be acceptable. This consumer might purchase both the foreign and domestic beers, drinking the foreign beer in limited amounts, say, on special occasions, and drinking 12Much of the information in this section on the attributes model is taken from Douglas (1987). 13For a review of beer demand estimates, see Ornstein (1980). Also see Heien and Pompelli (1988). The estimated Growth in the Demand f o r Foreign B eer— Increases in consumer incomes can also spur the demand fo r foreign products. W hen con sumers’ incomes increase, they are able to pur chase m ore o f all o f the products they desire. In general, how ever, the quantities purchased o f some goods, like flour, decline as incomes rise, w hile quantities o f other goods purchased, like furniture, increase as incomes rise. The statistical evidence relating beer consumption to income grow th is m ixed.13 Some studies have shown that the quantity o f beer consumed increases as income increases, w hile others have shown the opposite. Although little w ork has been done to estimate the relationship betw een foreign beer consump tion and income in general, there is some data to suggest that the demand fo r foreign beer might be positively influenced by increases in income. All 21 OECD countries in table 1 that provided im port data on beer had positive per capita gross domestic product grow th between 1975 and 1986; 16 o f these reported a positive average annual rate o f grow th o f beer imports per capita and imports as a percent o f consump tion. In addition, the market fo r im ported beers g rew much m ore rapidly than most domestic beer markets during the late 1970s and early 1980s, a period o f income grow th fo r most countries. These figures roughly suggest that per capita income grow th has contributed to the internationalization o f the brew ing industry. income elasticity of beer in these studies ranged from -0.46 to 0.79. NOVEMBER/DECEMBER 1990 10 Where to P rodu ce Beer: Some Econom ics Once a firm determines that foreign demand fo r its products exists, it must determine the lowest-cost method o f supplying these products to the foreign market. Should the firm use direct exports, a licensing agreem ent or direct foreign investment to enter the target market? The answer is relatively simple, in theory, and is based on the principle o f profit maximization. A brew ery's total cost o f supplying a foreign m arket is equal to the b eer’s cost o f production plus transportation and distribution costs, m ar keting costs and overhead. A b re w e ry ’s cost o f producing beer is a function o f its production technology and the cost o f its inputs, such as labor, agricultural ingredients and packaging materials. Research has shown that the average cost o f producing beer declines as production expands.14 That is, economies o f scale exist in the brew ing industry. Economies o f scale en courage direct exporting w hen the quantity demanded o f foreign beer is relatively small and encourage foreign production either through licensing or foreign direct investment w hen the quantity demanded is relatively large. Suppose a U.S. b re w e r and a Japanese b rew er have identical production functions exhibiting economies o f scale and that the firms pay the same price fo r their inputs. That is, their average cost o f production curves are equal and are shaped as shown in figure 2. (For simplicity, assume that the U.S.-Japan exchange rate is fixed throughout and, given this exchange rate, Japa nese prices are stated in U.S. dollars.) As the brew ers expand production, the average cost o f producing a unit o f the product falls up to a point, after which average costs no longer de cline but stabilize. Assume that Japanese de mand exists fo r a beer—called Colony—produced b y the U.S. firm. The U.S. firm must determine w hether it can supply the Japanese market cheaper b y producing Colony domestically and exporting or by producing it in Japan, either under license or by FDI. N ow suppose that supply and demand condi tions and price in the U.S. are such that 14Elzinga (1973), Fuss and Gupta (1981), Keithman (1978) and Scherer (1973) all provide evidence that economies of scale exist in the brewing industry. See Thompson (1985) or any micro-economic text for a discussion of the reasons why economies or diseconomies of scale can exist at the plant level. FEDERAL RESERVE BANK OF ST. LOUIS Am erican consumers consume Q us units o f Col ony, as shown in figure 2. (Note that the supply and demand curves are not shown in figure 2 and the quantity Q os is simply given.) This sub stantial amount o f consumption allows the U.S. firm to achieve significant economies o f scale, producing Q us units o f Colony at an average cost o f C2 per unit. Also assume that market conditions are such that a relatively small quan tity o f Colony, Q j, , is demanded in Japan. Since the U.S. b re w e r is already producing some Col ony fo r domestic consumption, expanding p ro duction to meet the extra demand o f Colony in Japan w ould allow the U.S. b re w e r to m ove dow n its average cost-of-production curve from point A to point B, w h ere it could produce Col ony fo r C,.15 The alternative to producing the beer in the United States and exporting it is to produce Col ony in Japan. Since Colony is currently not be ing produced in Japan, the Japanese firm or branch b re w e ry built by the U.S. firm would have to b rew the relatively small amount o f Col ony, Q ,, , at a high average cost o f production, C3. In other words, the relatively small quantity o f production w ill not allow the Japanese plants to achieve significant economies o f scale. Thus, producing Colony in the United States fo r ex port would save the b re w e r C3- C , per unit of Colony. If the cost o f transporting Colony to Japan and distributing Colony in Japan is less than the difference betw een C3 and C,, then the U.S. b rew er w ould maximize profits by exporting Colony to Japan. If the quantity demanded o f Colony in Japan w ere larger, it might be m ore profitable fo r the b re w e r to use a licensing agreement or foreign direct investment. Suppose that the quantity o f Colony sold in Japan grow s to Q ,2 as shown in figure 3, while sales o f Colony in the United States remain at Q us. Since the U.S. b re w e r has exhausted its economies o f scale, it cannot produce Q us+ Q ,2 at a low er per unit cost than that fo r Qus+Qj,. The Japanese brew ery, how ever, by increasing production from Q n to Q J2 could n ow match the U.S. b re w e r’s cost o f production because it has also achieved the low est possible average cost o f production. Thus, given equal average ,5For simplicity, we have ignored any quantity response, stemming from lower prices, that might occur in the U.S. market as a result of the expanded output. 11 Figure 2 Average Cost of Production: U.S. Plant Achieves Economies of Scale Figure 3 of output Average Cost of Production: U.S. and Japanese Plants Achieve Economies of Scale of output NOVEMBER/DECEMBER 1990 12 production costs, the U.S. firm w ill now prefer to either negotiate a licensing agreem ent with the Japanese b rew er or use FDI, thereby saving the additional export-related expenses o f ship ping Colony overseas and distributing it within Japan. Like transportation costs, trade barriers also offset production cost advantages. I f a target country has high tariffs or distribution systems fo r im ported goods that are relatively costly, production cost advantages in the home country may be offset and licensing and foreign direct investment becom e the only feasible methods o f entering the target market. As shown below, trade barriers have had a significant effect on the choice o f licensing agreements and foreign direct investment in the internationalization o f the brew ing industry. BARRIERS TO BEER TRADE Japan In Japan, tw o types o f barriers inhibit foreign beer from entering the country. The most significant o f these is the Japanese distribution system. Th e Japanese have a complex multi tiered system, comparable to the U.S. beer distribution system, in which beer moves from producer to w holesaler to consumer.16 In addi tion, Japan has little warehouse space, which means shipments are smaller and m ore frequent than in the United States. Both aspects o f the Japanese distribution system raise the cost o f distributing beer in Japan, relative to less com plex systems. Japan also charges a small customs duty on im ported beer. These factors raise the cost o f exporting beer to Japan and make licens ing agreements or foreign direct investment relatively m ore attractive methods o f selling beer in Japan. Canada A G A TT panel ruled in 1988 that specific practices o f the Canadian provincial govern ments discriminated against im ported b eer.17 Canadian trade barriers include discriminatory mark-ups at provincial liquor outlets and dif ferent marketing techniques fo r foreign beer, 16See VandeWater and Curley (1990). 17G. Heileman Brewing Co. filed a Section 301 trade action against Canada over unfair pricing and distribution prac http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS such as smaller packages and w arm foreign beer sales at the governm ental outlets. These non-tariff barriers have prom pted U.S. brew ers to use licensing agreements in Canada even though several b rew ers have U.S. plants that are located quite close to the Canadian border. The Australian b re w e r Elders IXL has chosen to use foreign direct investment to enter the Cana dian market. This creates an interesting situa tion in w hich a U.S. beer is being made under license in a Canadian b rew ery that is partially ow ned by an Australian brew er. United Kingdom As in Japan, distribution practices are the main barriers to trade in the United Kingdom. Most beer consumed in the United Kingdom is draft beer, and most o f this is sold in pubs. Many pubs are ow ned outright by breweries, managed b y the brew eries or leased to in dividuals w ho enter into exclusive supply agree ments w ith the breweries. This system was the subject o f eight investigations betw een 1966 and 1986, that focused chiefly on pricing and supply competition.18 Given the relationship betw een the pubs and the domestic brew eries, foreign label brew ers have problems getting local brew ers to carry their products in British pubs. Thus, many foreign brew ers have chosen to use licensing agreements with domestic firms to penetrate the U.K. beer market. Foreign direct investment has also been used to enter this market. SUMMARY The brew ing industry has evolved from an in dustry that concentrated on domestic markets to one that views itself as part o f a global m ar ket. This internationalization has occurred via the use o f merchandise trade, licensing agree ments and foreign direct investment. M erchan dise trade in beer has developed in an intra industry pattern, whereas international transac tions in licensing agreements and foreign direct investment have not developed, in general, in a bilateral pattern. Licensed production and p ro duction at foreign-ow ned brew eries likely ac count fo r an unknown, but probably large, part o f foreign beer consumption in some countries. tices of provincial governments. See Daily Report for Ex ecutives (1990). 18Brewers Association of Canada (1989), pp. 387-88. 13 T w o conditions must hold fo r trade in similar goods, such as beer, to occur. First, the foreign product must o ffer a combination o f desired at tributes that are not available to domestic con sumers from domestic products. Second, it must be profitable to produce the product fo r the foreign market. W hich type o f international transaction will be used to supply a foreign m ar ket with beer is related to the existence o f eco nomies o f scale, distribution costs and trade barriers. REFERENCES Brewers Association of Canada. International Survey: Alcoholic Beverage Taxation and Control Policies, 7th ed. (Ottawa, Canada, June 1989). Carrington, Tim. “ Britain Blocks Elders Takeover of Big Brewer; Government Urges Reforms in Reflecting Firm’s Offer for Scottish & Newcastle,” The Wall Street Journal, March 22, 1989. Carter, Colin A., Jeffrey Karrenbrock and William W. Wilson. “ Freer Trade In the North American Beer and Flour Markets,” in Andrew Schmitz, ed., Free Trade and Agricultural Diversification: Canada and the United States, (Westview Press, 1989). The Conference Board of Canada. The Canadian Brewing In dustry: Historical Evolution and Competitive Structure, (Ot tawa, Canada, February 1989). Daily Report for Executives. (The Bureau of National Affairs, Inc., Washington D.C., May 18, 1990), p. A-3. Douglas, Evan J. Managerial Economics: Analysis and Strategy, 3rd ed. (Prentice-Hall, Inc., 1987), pp. 86-103. Elzinga, Kenneth G. “ The Restructuring of the U.S. Brewing Industry,” Industrial Organization Review (Vol. 1, 1973), pp. 101-14. Finnegan, Terri. “ International Licensing Pacts on the Rise,” Modem Brewery Age (July 13, 1987). Food and Agricultural Organization of the United Nations. FAO Trade Yearbook, various issues. Fuss, Melvyn A., and Vinod K. Gupta. “A Cost Function Ap proach to the Estimation of Minimum Efficient Scale, Returns to Scale, and Suboptimal Capacity,” European Economic Review (February 1981), pp. 123-35. Gray, H. Peter. International Economic Problems and Policies (St. Martin’s Press, Inc., 1987). Harris, Clay. “ Time for Takeovers, Gentlemen Please,” Financial Times, March 14, 1990, London, England. Heien, Dale, and Greg Pompelli. “The Demand for Alcoholic Beverages: Economic and Demographic Effects,” Southern Economic Journal (January 1989), pp. 759-70. Horstmann, Ignatius, and James R. Markusen. “ Licensing Versus Direct Investment: A Model of Internationalization by the Multinational Enterprise,” Canadian Journal of Economics (August 1987), pp. 464-81. International Monetary Fund. International Financial Statistics Yearbook (1989), pp. 122-23. “Japan’s Asahi Plans Brewery in U.S.” St. Louis PostDispatch, January 3, 1990. Keithman, Charles. The Brewing Industry, Staff Report of the Bureau of Economics, U.S. Federal Trade Commission (1978). Lancaster, K. "A New Approach to Consumer Theory,” Jour nal of Political Economy (April 1966), pp. 132-57. _______ . Consumer Demand: A New Approach (Columbia University Press, 1971). Ornstein, Stanley I. “Control of Alcohol Consumption Through Price Increases,” Journal of Studies on Alcohol (September 1980), pp. 807-18. Ott, Mack. “ Have U.S. Exports Been Larger Than Reported?” this Review (September/October 1988), pp. 3-23. Scherer, F.W. “ The Determinants of Industrial Plant Sizes in Six Nations,” Review of Economics and Statistics (May 1973), pp. 135-45. Sherwell, Chris. “ Foster’s Brewing Sees Global Sales of A$9bn,” Financial Times, March 14, 1990, London, England. Thompson, Arthur A. Economics of the Firm: Theory and Practice, 4th ed. (Prentice-Hall, Inc. 1985). Thornhill, John, and Clay Harris. “GrandMet and Elders Un veil Swap Details,” Financial Times, March 14, 1990, Lon don, England. VandeWater, Judith, and John Curley. “ Beer Barrels Eastward,” St. Louis Post-Dispatch, June 11, 1990. Weinberg, R.S. & Associates. “Tax Free Removals, Destina tion of Exports and Consumed on Brewery Premises, 1947-1988” (Brewing Industry Research Program, St. Louis, Missouri). NOVEMBER/DECEMBER 1990 14 Appendix A Foreign Demand fo r Beer: The Attributes Model The attributes model, introduced by Lancaster (1966, 1971), can be used to show how demand fo r foreign beer can exist in a country in which domestic brands are already produced.1 Suppose a consumer chooses betw een tw o brands o f beer so that, subject to income limitations, his satisfaction from the services provided by the beer is maximized. For simplicity, assume the consumer values only tw o attributes o f beer: taste and low calories. The tw o types o f beer provide these attributes in differing proportions and at different prices. Figure A1 Maximization of Utility by Consuming Only Domestic Beer A fter sampling both products, the consumer rates each brand on a scale o f 1 to 3, 3 being best, fo r both taste and calorie content, as shown in table A l. Brand F is a foreign beer that tastes great, but is high in calories (thus receiving a low rating on caloric content) giving it a rela tively high ratio o f taste-to-calorie appeal. Brand D, a domestic beer, does not taste quite as good, but is very low in calories. This beer then has a relatively low taste-to-calorie appeal ratio. Table A1 Attribute Ratings and Prices of Three Beers____________________________ Attribute Rating Brand Calories Taste Ratio of taste to Bar price Bottles calories per bottle per $12 D 3 1 1/3 $2 6 F 1 3 3 $3 4 The amount o f each beer the consumer can purchase is determined by his income and the price o f the products. Assume that the consumer has decided to spend $12 on beer during a visit to a local bar and the cost of each kind o f beer is as shown in table A l. If he spends the entire $12 on only one p ro duct, he could buy at most six bottles o f D or four bottles o f F. Four bottles o f F would provide 4 units (4 bottles X 1 unit o f caloric 1Much of the information in this appendix follows the discussion of the attributes model as presented by Douglas (1987), pp. 86-102. FEDERAL RESERVE BANK OF ST. LOUIS appeal per bottle) o f the caloric attribute and 12 units (4 X 3) o f the taste attribute. The tw o products are depicted in figure A l in an attribute space as rays from the origin. The slope o f each ray is determined by the ratio of taste to calorie appeal. If the consumer drinks brand F, then he moves out along ray F, absorb ing the tw o attributes in a ratio o f 3:1. Points A and B represent the maximum amount o f the tw o attributes that can be obtained by consum ing beers F and D, respectively, given the spend ing constraint o f $12. Joining points A and B provides the consumer's efficiency frontier. The efficiency fron tier is the outer boundary o f the attainable combination of the tw o attributes, given the budget constraint o f $12. It is called efficient because a utilitymaximizing consumer w ill get m ore utility by being on the fron tier rather than within the frontier, even though these interior points are attainable. 15 Figure A2 Figure A3 Maximization of Utility by Consuming Only Foreign Beer Maximization of Utility by Consuming Both Domestic and Foreign Beer H ow do w e know which beer, or combination o f beers, the consumer w ill choose? In the at tributes model, consumer preferences betw een attributes can be expressed using indifference curves. Like indifference curves used to express the marginal rate o f substitution (MRS) betw een tw o products, the attribute indifference curves express the MRS betw een attributes, and higher indifference curves represent higher levels o f utility. The beer consumer's assumed attribute indifference curves I, and I2 have been superim posed on the attribute space in figure A l. As suming that a consumer wishes to maximize his satisfaction from taste and caloric content, he w ould choose to be on the highest attainable in difference curve, which occurs at point B in figure A l. The position and slope o f the indifference curves w ill determine the brand or brands o f beer chosen. This particular consumer has an indifference curve that is relatively steep indi cating that, compared with a consumer with a flat indifference curve, he is willing to give up a lot o f taste to get a few less calories. the foreign brand F, at point A. Thus, in a socie ty w h ere some consumers p refer the attributes o f foreign beers over domestic beers, a demand fo r foreign beer w ill exist. Figure A3 shows an example o f a consumer w ho would purchase both foreign and some domestic beer. Neither o f the beers provides the attributes exactly in the ratio represented by point N. The consumer could reach this point, how ever, by consuming some o f both products. Bv consuming L units o f the domestic brand, the consumer would obtain Y, units o f taste and X ( units o f caloric appeal. By spending the rest o f his budget on brand F, the consumer would travel along the path LN, which has the same slope as ray F, to obtain the X2-X , units o f calo rie appeal and the Y 2- Y , units o f taste needed to reach his maximum level o f utility at point N. Alternatively, the consumer could have started by consuming M units o f brand F and then con sumed L units o f brand D to reach point N. Growth in Foreign Demand N ow suppose that a different consumer, w ho gives the beers the same attribute ratings, is willing to consume a lot m ore calories to gain a bit better taste. The shape o f this consumer’s in difference curve w ould be m ore flat, and as shown in figure A2, this person would choose The demand fo r foreign beer can increase if the relative price o f the foreign beer falls. As shown in figure A4, w hen the price o f the fo r eign beer falls, the maximum amount o f the foreign brand that can be purchased increases, NOVEMBER/DECEMBER 1990 16 Figure A4 Figure A5 The Price Effect Shown by the Attribute Approach The Income Effect Shown by the Attribute Approach shifting out the efficiency fron tier from AB to BC. In this example, the consumer goes from buying some o f both brands at point X to only buying the foreign brand at point C. consumption o f each beer. T o maximize utility, the consumer shifts to point C on the higher in difference curve I2, w h ere he consumes some o f both brands o f beer. Thus, a higher level o f in come has induced the consumer to purchase m ore o f the foreign brand o f beer. Expanding incomes can also explain increased consumer demand fo r foreign beer. In figure A5, the consum er’s increase in income has led to a shift from buying only the domestic beer to buying some o f both beers. Initially the con sumer’s efficiency fron tier is AB, the highest in difference curve attainable is I t and the con sumer purchases only the domestic product D. W hen the consumer’s income increases, the e ffi ciency frontier shifts out in a parallel fashion to A'B', because prices and attributes are fixed and only income is changing. For the given prices and attributes, the increased income allows m ore FEDERAL RESERVE BANK OF ST. LOUIS O f course, the example could have been con structed to show how a shift in the consumer’s income could have led to a reduction in the amount o f foreign beer purchased. Several re searchers have estimated the demand fo r beer, but no conclusive evidence has been provided as to w hether b eer consumption expands as a consumer’s income expands. Some data, as dis cussed in the text, how ever, suggest that the de mand fo r foreign beer might be positively in fluenced by increases in income. 17 Appendix B Whether to Produce Domestically or Abroad: Some Economics Assume that the long-run average cost (LRAC) curves o f any plants w h ere Colony could be b rew ed are identical. These potential plants in clude the U.S. parent firm and any branch plant it may establish in Japan, as w ell as any existing b rew ery in Japan that could b rew Colony under license. In addition, assume that the U.S.-Japan exchange rate is fixed throughout and, given this exchange rate, Japanese prices are stated in U.S. dollars. in figure Bl, part o f the demand curve is above the LRAC curve, indicating that consumers are willing to pay a price fo r Colony that is above the U.S. firm 's average cost fo r some levels o f production. In Japan, how ever, if the demand fo r Colony is relatively small, say D ,, as shown in figure B2, then Colony cannot be b rew ed profitably there. At each quantity along D „ , the price consumers are willing to pay is below the Japanese plant's average cost o f producing Col ony. Thus, when demand is D,,, direct exports from the U.S. b re w e ry will be the only w ay the Japanese market might be supplied with Colony. However, if the Japanese demand schedule for Colony w ere much larger, say, at D J2, then the Japanese b rew ery could at least cover its p ro duction costs fo r some level o f production. Thus, at this larger level o f demand, both p ro duction in the United States and in Japan are potential routes o f supplying Japan with Colony. Consider the level o f demand fo r Colony in the tw o countries. In the United States, as shown Consider the cost o f producing different levels o f Colony fo r foreign consumption m ore closely. Suppose that a U.S. b rew er wants to sell its beer, Colony, in Japan. The U.S. b rew er has three methods o f supplying Japan with Colony: 1) produce Colony domestically and export to Japan, 2) negotiate a licensing agreement w ith a Japanese b rew er w h o w ould b rew and distribute Colony in Japan, or 3) produce Colony in Japan either by purchasing an existing Japanese b re w ery or building a new brew ery. W hich is the cheapest? Figure B1 Figure B2 U.S. Production Costs and Demand for Beer Japanese Production Costs and Two Demand Curves for Beer NOVEMBER/DECEMBER 1990 18 Figure B3 Figure B4 U.S. Plant’s Production Costs Japanese Plant’s Production Costs Compared to U.S. Plant’s Cost of Exporting The goal is to determ ine which type o f interna tional transaction allows the firm to provide Colony to the Japanese market at the lowest cost. First, consider the U.S. b re w e r’s cost of producing and exporting Colony to Japan. As sume fo r a given market price, the quantity o f Colony demanded in the United States is Q us, as shown in figure B3. The U.S. b rew er produces this amount at an average cost o f C4 per unit. U.S. brew er's cost o f producing different amounts o f Colony fo r export to Japan, shown by the line LR A C *S in figure B4. N ow suppose that fo r a given price, the quan tity demanded o f Colony in Japan is Q,,, as shown in Figures B3 and B4. Since the U.S. b rew er already produces some Colony fo r do mestic consumption, by expanding production to Q us + Q ji to meet the export demand, the U.S. b re w e ry could m ove down its LRAC curve from point A to B in figure B3, low ering its average cost o f producing Colony from C4 to C^.1 If the quantity demanded o f Colony in Japan w as larger, at Q J2, the U.S. b re w e ry ’s average cost o f producing it w ould fall even further to C, at point C. By doing a similar analysis fo r other quantities o f U.S. exports, w e can develop the Figure B4 allows a straightforw ard com parison o f the production costs o f exports to Japan, LRAC£,S, with the production o f these quantities in Japan, LRACj. It shows that the average cost faced by the U.S. b rew er produc ing a given amount o f Colony fo r export is low er than the Japanese b re w e r’s average cost, LRAC „ up to the quantity Q J3, but higher fo r all subse quent levels. This is possible, even when cost curves are identical across countries, because the U.S. plant was already producing Colony fo r domestic consumption and that by expanding production to meet export demand, the average cost o f producing Colony fell. The Japanese plant currently is not producing any Colony; if it w ere to start brew ing Colony fo r Japanese consumption, it w ould have to start at a higher cost on its long-run average cost curve. Economies o f scale, how ever, do not continue indefinitely. Consequently, the cost o f producing 1For simplicity, the analysis ignores any potential sales price decline in the United States that may stem from the brewer achieving greater economies of scale. Thus, the quantity demanded in the United States, after production is increased to meet export demand, is assumed to remain at the same level as before export production occurred. FEDERAL RESERVE BANK OF ST. LOUIS 19 Colony eventually starts to rise and the Japanese plant can produce Colony cheaper than the U.S. plant after point Q J3. Thus far, the discussion has focused solely on the cost o f producing Colony. Transportation and distribution costs are likely to influence w h ere production is located. Assume that it costs the same per unit to ship and distribute a small amount o f Colony as it does a large amount o f Colony. Since the U.S. plant has to ship Colony overseas, it is reasonable to assume that its transportation and distribution costs w ill be sig nificantly higher than a Japanese plant’s w ould be if Colony w ere produced there. Adding these average per unit transportation and distribution costs to the plant's respective long-run average cost o f production curves gives the tw o dashed lines, LR A C *s+ t us and LRACj + t,, shown in figure B4. The U.S. plant can produce, transport and distribute Colony to the Japanese market at a low er cost than the Japanese b rew ery can up to the quantity Q*. Notice that the additional costs o f transportation and distribution have low ered the quantity at which the U.S. b rew ery can compete from Q J3 to Q*. At quantities beyond Q*, the Japanese firm can produce and distribute Colony fo r less than C3, giving it a cost advantage over the U.S. b rew er.2 Because o f the cost advantage, fo r any given quantity o f Colony demanded in Japan up to Q*, the U.S. firm would p refer to produce Colony domestically and export the product to Japan. If the quantity o f Colony demanded in Japan w ere greater than Q*, the U.S. firm w ould either at tempt to negotiate a licensing agreement with the Japanese b rew er or purchase or build a Japanese b rew ery fo r production o f Colony. W hether the U.S. b rew er w ould choose to license production or open a branch b rew ery in Japan w ould depend on several factors. Horstmann and Markusen (1987) note that if the li censee and the plant to be built or purchased are equally efficient, then the need to give the licensee the incentive to maintain the reputation o f the licenser’s product w ill result in FDI always dominating the use o f licensing. They also conclude, how ever, that if the licensee and branch plant are not equally efficient—that is, if their LRAC curves are not identical—then other factors such as the size o f the market, the ex istence o f close substitutes in the target market and the level o f interest rates in the tw o coun tries will determine w hether licensing agree ments or FDI w ill be used. In addition to production and distribution costs, brew ers also face ta riff and non-tariff trade bar riers, which raise the cost o f supplying a coun try with beer. In terms o f figure B4, it is con ceivable that the U.S. b rew ery could have an average cost o f production considerably below brew eries in Japan, but that trade barriers in Japan are so high that licensing agreements or foreign direct investment becom e the preferred method o f supplying the foreign country at all levels o f demand. Here, the LR A C *s+ 1us curve can be used to incorporate this idea. Let the t us variable now stand fo r transportation and distribution as w ell as costs associated with trade barriers, such as tariffs. The existence o f trade barriers simply shifts the U.S. b re w e r’s export cost curve upward, pushing Q* closer to the origin. Other realistic problems associated with inter national transactions have been ignored in this example. Some o f the other factors that would affect how a firm supplies a foreign market in clude differences in production technology and input costs, governm ent restrictions on foreign investment, costs o f negotiating and monitoring licensing agreements, exchange-rate movements and the role other products being produced at the brew eries might have on the plant’s cost o f production. 2Of course, the Japanese firm will eventually reach its points of diseconomies of scale and its average cost of production will rise above Cr NOVEMBER/DECEMBER 1990 20 Michael T. Belongia and James A. Chalfant Michael T. Belongia is an assistant vice president at the Federal Reserve Bank of St. Louis. James A. Chalfant is an associate professor of agricultural and resource economics at the University of California-Berkeley. Scott Leitz, Lora Holman and Lynn Dietrich provided research assistance. Alternative Measures of Money as Indicators of Infla tion: A Survey and Some New Evidence I n OCTOBER 1982, the Federal Open Market Committee (FOMC) officially de-emphasized M l as an interm ediate target variable in conducting m onetary policy. Th e FOMC further reduced the emphasis given to M l in 1987 by not setting target ranges fo r its rate o f growth. The typical explanations given fo r this de-emphasis have stressed a breakdow n in the 1980s in the rela tionships betw een M l grow th and both the rate o f inflation and the grow th rate o f nominal GNP that had characterized the post-war period.1 The de-emphasis o f M l as a policy guide has prom pted many observers to seek a replace ment. Some analysts have argued that financial 1Specifically, in the postwar period through 1981, M1 growth was related closely to nominal GNP growth over a period of one year and to the rate of inflation over a period of three to five years. The breakdown in these relation ships also has been discussed as a decline in the growth rate of M1 velocity. See Stone and Thornton (1987), for ex ample, for a survey of explanations that have been offered for this abrupt shift in what had been quite stable relation ships. Tatom (1988) and others, in contrast, have disputed http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis innovations and deregulation have altered the relationship betw een interest-bearing and noninterest-bearing assets. One group o f individuals, fo r example, has argued that the interest-bearing components o f M2 are good substitutes fo r the narrow er set o f assets in M l and, on this basis, concluded that M2 now is the best m onetary ag gregate fo r policy actions. A nother group, rea soning that interest-bearing accounts act m ore as savings balances than transaction accounts, has advocated the narrow measure o f M IA (cur rency plus non-interest-bearing checkable depos its).2 Still others have advocated the use o f weighted monetary aggregates, arguing that the individual assets included in M l, M2 and the whether historical relationships between M1 and nominal variables were fundamentally different in the 1980s. 2For the arguments favoring M2, see Hetzel (1988). For the evidence favoring M1A, see Hafer (1984) and Darby, et al. (1989). 21 broader aggregates possess different degrees o f "moneyness” according to the interest they pay. If these assets have different degrees o f m oney ness, they should be weighted differently when added together.3 This article focuses on research that addresses both the composition and w eighting issues in deriving a new monetary aggregate from the first principles o f demand analysis.4 It surveys this w ork and evaluates the perform ance o f tw o new money stock measures as indicators o f future inflation. W H Y IS THE DEFINITION OF “MONEY” IMPORTANT? W hat truly constitutes "m oney” in the real w orld is an interesting and important issue because, as economic theory predicts, money has pow erfu l effects on economic activity. Un fortunately, theory alone does not o ffe r explicit guidance in choosing a unique and unambig uous real w orld counterpart to the abstract theoretical concept o f "m oney.” W ithout clear theoretical guidance, many economists have adopted the conclusion that the appropriate definition o f money is the one that serves best in the chosen empirical application.5 One theoretical fram ew ork that attributes an important causal role to m oney in determining fluctuations in economic activity is the Quantity Th eory o f Money. This theory is based on the Equation o f Exchange, which typically is w rit ten as: (1) M V = PT, w h ere M is the quantity o f money, V is its velocity o f circulation, P is the average price level and T is the volume o f transactions. The Quantity Th eory derived from this equation predicts that the price level varies directly with the quantity o f money. In its strictest form, which has been subjected to empirical testing, the Quantity Th eory predicts that the causation runs from changes in m oney to changes in prices (but not the reverse) and that, accounting 3See, for example, Barnett (1983), Barnett et al. (1984) or Spindt (1985). 4See Belongia and Chalfant (1989) and Swofford and Whitney (1990a,b). 5See Friedman and Schwartz (1970), p. 91. fo r the short-run influences o f other variables, changes in m oney and prices are proportional.6 Obviously, the definition o f m oney is im por tant in testing such propositions. The definition o f m oney is also important fo r m onetary policy purposes. Assume, fo r example, that a central bank believes the implications o f the Quantity Theory. It might then direct its policies to achieve a grow th rate o f m oney consistent with zero inflation. This raises the issue o f how to measure m oney in the real w orld. As Laidler (1989) points out, both critics and advocates o f the Quantity Th eory frequently disagree about the appropriate definition o f m oney to test its implications. Thus, a central bank committed to achieving a policy goal fo r "P ” still faces con siderable uncertainty about which "M ” should be targeted and how to do so. FINANCIAL INNOVATIONS AND THE CONDUCT OF MONETARY POLICY For m ore than 30 years after W orld W a r II, the trend grow th rate o f M l velocity was nearly constant at 3 percent. For many economists, this stability made M l grow th a reliable leading indicator o f the effects o f current m onetary ac tions on future econom ic activity. As figure 1 shows, how ever, this seemingly stable trend was disrupted sharply in 1981, perhaps because o f the nationwide introduction o f interestbearing checkable deposits in January o f that year. Because these deposits not only paid rates o f interest that w e re close to those on savings deposits but also allowed depositors to w rite checks directly on these deposits, the hypothesis was that at least some portion o f NO W account balances w ere savings. If so, these new ac counts combined “ idle” savings balances with "active” transaction balances in traditional de mand deposits. The larger quantity o f balances in M l accounts, fo r the same level o f GNP, w ould be reflected in a fall in the grow th rate o f M l velocity.7 M ore important, uncertainty about the effects o f financial innovations on M l, 6See Laidler (1989) for a thorough review of the Quantity Theory’s history, implications and controversies. 7See Gordon (1984), for example, for more discussion of this argument. NOVEMBER/DECEMBER 1990 22 Figure 1 Log of M1 Velocity1 'Velocity data are indexed to a base value of 100 in 1/1963. in addition to velocity growth, weakened M l ’s informational content in predicting future price movements.8 This uncertainty has been an important ele ment o f m onetary policy discussions. The record o f the FOMC, fo r example, shows that, at the meeting o f February 8-9, 1983: Committee members’ views varied considerably on the weight to attach to M l ... the perfor mance of that aggregate had been subject over the past year or more to substantial uncertain ties related to the growing role of NOW ac counts and an apparent shift in the behavior of its income velocity ... Only modest allowance was made for the possibility that the new Super NOW accounts would draw funds into M l from other sources.9 8Monetary control also will be impaired if shifts of savingstype balances into interest-bearing checkables weakens the link between the monetary base and M1. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis T w o months later, in a statement to Congress, Federal Reserve Chairman Paul Volcker reported: To some extent — but it cannot be measured with any degree of certainty — the decreases in “velocity” may reflect the changing nature of M l; with interest-bearing NOW and Super NOW accounts making up an increasingly large proportion of Ml, this aggregate may be in fluenced by "savings" behavior as well as "tran sactions” motives. That is a longer-term factor, and the growth in M l over the shorter-run may have been affected by the reduced level of market rates — particularly relative to interestbearing NOW accounts — and slowing inflation as well. The range o f uncertainty on these points is substantial (emphasis added) and has led the Federal Open Market Committee to place less emphasis on M l in the implementation of policy over the short term.10 9Federal Reserve Bulletin, April 1983, pp. 288-289 10Federal Reserve Bulletin, May 1983, p. 338. 23 Nonetheless, after being “ m onitored” but not targeted in 1983, M l was chosen as an interm e diate target o f m onetary policy from 1984 to 1986. In 1987, the FOMC again declined to set target ranges fo r M l growth, instead choosing to emphasize M2 and M3 in policy discussions.11 These official statements point to considerable uncertainty about how interest-bearing checking accounts w ere being used and how their use might affect the behavior o f the m onetary ag gregates generally, the behavior o f M l specifi cally, and the relationship o f M l to important economic variables, such as spending and prices. APPROACHES TO THE DERIVATION OF ALTERNATIVE MONETARY AGGREGATES Choosing a m onetary aggregate requires at least four steps.12 First, the group o f assets to be included in the aggregate must be selected. Second, individual asset categories must be ag gregated appropriately. Third, the aggregate must be analyzed in terms o f the central bank’s ability to control it. Finally, its perform ance in some ultimate use must som ehow be evaluated. Before the 1980s, those w ho w ere skeptical o f using the M l aggregate fo r policy analysis gen erally focused on the fourth point, the m ethod ology used to link M l either to nominal GNP grow th or the inflation rate. Many critics, fo r example, raised issues about the exogeneity o f changes in the m oney stock, reverse causation running from income or prices to m oney or technical problems associated w ith estimation o f reduced-form equations.13 W h ile such m ethodo logical criticisms continued throughout the 1980s, much m ore attention in recent years has been focused on the aggregates themselves rather than their final uses. The official money stock measures particularly have been criticized both fo r the w eighting scheme used to ag- 11For more detail on the uncertainty about the behavior of different aggregates, see Hafer and Haslag (1988) or Gar finkel (1990). 12Barnett, et al. (forthcoming) argue that this process also involves addressing a fifth step—monitoring how closely the index number tracks the economic aggregator function—that has been overlooked by previous research in this area. 13See, for example, de Leeuw and Kalchbrenner (1969), Davis (1969), Gramlich (1971), Goldfeld and Blinder (1972) and B. Friedman (1977), among others. 14The full quote is, “ Money never bears interest except in the sense of creating convenience in the process of ex change.” See Fisher (1963), p. 9. gregate individual asset categories and fo r the specific assets used to construct an aggregate. These criticisms are discussed in m ore detail below. Composition o f Monetary Aggregates Historically, debates about w hat is m oney have focused on the things—coins, bank notes, etc.—that should be included in a m onetary ag gregate. Before the em ergence o f interestbearing checkable deposits, the logic behind the assets included in M l coincided w ith Irving Fisher’s dictum, “M oney never bears interest... ,"14 For economists o f this school, the m onetary ag gregate also was view ed as containing only those assets that could be readily exchanged fo r goods.15 Since the birth o f m oney market mutual funds (MMMFs) and, even m ore so, the nation w ide introduction o f N O W accounts in January 1981, how ever, the question o f which assets are correctly identified as m oney has becom e com plicated . Many argue that the historical justifi cation fo r the group o f assets in M l is no longer clear. What, fo r example, distinguishes the NOW accounts in M l from MMDAs or MMMFs, which appear to have characteristics similar to NOWs but are excluded from M l? Conversely, because N O W accounts pay interest and can be used as savings accounts, should they be excluded from a m onetary aggregate designed to include those assets that can be readily exchanged fo r goods? Tests f o r Weak Separability and Asset Selection One approach to deciding which financial assets should be considered as a group fo r ag- 15lbid., p. 8. NOVEMBER/DECEMBER 1990 24 gregation treats financial assets as commodities that are held fo r the services they provide. In this approach, individuals are assumed to allocate their financial wealth across the spec trum o f available assets, according to their preferences fo r the characteristics o f each asset and the relative returns each asset yields. Treating financial assets in this manner leads to inserting them as arguments in the utility func tion o f the representative individual and analyz ing the demand fo r financial assets in much the same w ay that the demand fo r other com modities can be studied.16 This approach leads to a search fo r “ separa ble” com m odity groups. Separability is related to the notion o f multi-stage budgeting in which consumers first allocate their budget across broad expenditure categories such as “ shelter,” "transportation,” “ clothing” and “food.” Then after allocating, say, 20 percent o f their total ex penditures to food, further allocations are made within the food group as expenditures are bud geted fo r "m eat,” "dairy products,” “ fruits and vegetables” and so on. The process continues w ith the gross expenditures fo r meat allocated across beef, pork and chicken. The result o f this process can be pictured as a branching tree diagram in w hich broad commodity groups are continually divided into smaller, m ore specific com m odity groups. T o decide w hich financial assets may be grouped under the heading "m oney,” the re search task is to find what goods are in the “ m oney" group and what goods are unrelated to utility from the services o f m onetary assets. This identification can be done form ally by testing w hether an asset collection is w eakly separable from other assets and commodities. Indeed, Barnett (1982, pp. 695-96), has shown that a necessary condition fo r the existence o f a m onetary aggregate is that the subset o f m one tary goods is w eakly separable from nonmone tary goods.17 16There has been a steady controversy about whether money can be modeled as an argument in the utility (or production) function; however, Feenstra (1986) has shown that this approach is equivalent to other standard ap proaches, including the Clower (transactions) constraint; see Clower (1970). 17For more detail on weak separability and other points related to finding an admissible monetary aggregate, see Varian (1982, 1983), Barnett (1987), Ishida (1984), Serletis (1987), Fayyad (1986), Hancock (1987) and Belongia and Chalfant (1989). FEDERAL RESERVE BANK OF ST. LOUIS W eak separability, a result from demand theory, implies that the consum er’s marginal rate o f substitution betw een tw o assets in a group is unaffected b y the quantities o f any good not in the group. This ensures that changes in the quantities consumed o f goods not in the group w ill not affect the marginal utilities de rived from goods within the group. The practi cal implication o f w eak separability is that it permits demand analysis that is concerned only w ith a specific collection o f assets; it does not re quire that one study all goods in the economy. T o view the problem differently, let some group o f assets, A, consist o f goods a l, a2 and a3; that is, A = (ai, a2, a3). Assume, how ever, that this group is not w eakly separable from other assets; instead, the separable group also should include asset a4. By arbitrarily cutting o ff the group to include only al-a3, changes in the quantity o f a4 consumed w ould affect the marginal rates o f substitution w ithin the group o f al-a3 and give rise to seemingly unstable preferences fo r asset holdings or, from a p ro duction function point o f view, an unstable transactions technology. In a m ore concrete sense related to traditional empirical w ork, the omitted variable a4 w ould affect an estimated demand function fo r an aggregate variable con structed from assets al-a3 b y acting as an om it ted shift variable. For example, estimating a m oney demand function based on M l balances when MM DAs should be part o f this group could be the source o f instability in such a de mand function.18 Recent w ork by S w offord and W hitney (1987, 1988) and by Belongia and Chalfant (1989) has examined the conditions o f w eak separability fo r the asset groups that correspond to M l and M IA (M l minus interest-bearing checkable de posits), as w ell as broader collections o f assets that include m ore components o f M2. This re search checked fo r the existence o f a well-behaved utility function that was consistent with 18ln fact, studies of money demand functions often include a dummy variable around 1981-82 to capture the effects of financial innovation. Aggregates based on weakly sep arable groups, however, should be stable and eliminate the need to account for shifts in functional relationships with dummy variables. 25 the financial balances held in various form s at existing prices and levels o f income. Upon finding that the data w ere consistent w ith demand theory, the separability tests w ere then used to identify the specific financial assets, in combin ation w ith currency, that w ere w eakly separable from other financial assets in the utility func tion. These w eakly separable asset groups w ere then candidates fo r aggregation. Weighting o f Asset Components Another criticism o f M l as an indicator o f m onetary actions—specifically w hen examining w h y the financial innovations o f the early 1980s might be associated with the sharp slowing in M l velocity g row th —focuses on the potential e r rors associated w ith the simple-sum weighting scheme used to derive M l and other official m oney stock measures. The simple-sum aggrega tion scheme, which gives the same w eight to currency as it does to dollars deposited in ac counts with quite different characteristics, violates fundamental aggregation principles, so the argument goes. For example, a dollar o f currency (which is term ed pure m oney because it pays no explicit rate o f return) is weighted equally in M l w ith a dollar in a N O W account deposit. This weighting occurs even though the N O W account pays a market rate o f return that, in recent years, has approximated the return on savings deposits. Thus, even though some N O W account balances may be held as savings, they are included dollar fo r dollar in M l w ith the other assets that tradi tionally have been considered to be transactions balances.19 Such equal weights are justified only if the assets are perfect substitutes fo r each other; both the raw data as w ell as much em pirical w ork demonstrate that these conditions do not hold.20 19Some analysts, in fact, have attempted to infer the proportion of NOW balances held for that asset’s savings characteristics (its interest rate and deposit insurance) on the basis of turnover data relative to demand deposit turn over rates. See Spindt (1985). 20See Barnett (1982) for the technical arguments about the shortcomings of simple-sum weighting. Ewis and Fisher (1984, 1985) and references they cite are sources that find low elasticities of substitution between M1 components and near-monies. 21See, for example, Friedman and Schwartz (1970), pp. 151-52. The fundamental point here is that each dollar in a N O W account should receive less w eight than a dollar o f currency, which p re sumably is held purely fo r the conduct o f trans actions. Similarly, if the assets collected in M2 are found to be a w eakly separable group, each o f its components should be w eighted different ly w ith the weights declining as own-rates o f in terest rise. A t this point, it should be noted that, w hile simple-sum weighting is clearly w ron g in principle, it may not in the past have m attered empirically; that is, it may not have produced major distortions in the estimated relationships betw een "m oney” and other vari ables. It also should be noted that the interest in w eighted m onetary aggregates is not new; the deficiencies o f simple-sum aggregates w ere recognized m ore than 20 years ago.21 By form alizing this intuitive argument, Barnett and others constructed a series o f Divisia m one tary aggregates.22 A Divisia index, w hich is m erely one o f many statistical index number fo r mulas that have desirable aggregation prop er ties, has an advantage over simple-sum aggrega tion in that it internalizes pure substitution e f fects. This means that the measured value o f the index w ill not change unless the utility or production functions underlying the index actu ally produce a different level o f utility or output. In contrast, the failure o f simple-sum weighting to internalize substitution effects means it can change even w hen there has been no change in the flo w o f transaction services from the group o f assets. Unlike the simple-sum weighting scheme cur rently used, any good index gives different weights to balances held in different deposit categories. In the case o f the Divisia index ap plied in m onetary studies, the weights are based essentially on the difference betw een the rate o f return on a pure store o f value (typically, a long-term bond) and each asset’s ow n rate o f 22See Barnett (1983, 1984) and Barnett, Offenbacher and Spindt (1981, 1984). NOVEMBER/DECEMBER 1990 26 return.23 The larger this interest rate differen tial, the higher the opportunity cost o f holding a particular financial asset and the greater its "moneyness.” The largest interest differential and, therefore, the greatest "moneyness” occurs fo r currency, whose rate o f return is zero. The research cited above has put the defini tion and measurement o f m onetary aggregates on firm er economic and statistical footing; the specific aggregates it has produced, how ever, have been subject to criticism. One criticism has been that w eighting may be, or at least may have been, unimportant empirically. Specifically, the relative perform ance o f simple-sum vs. Divisia measures against various final uses cri teria has been disputed.24 Perhaps the most im portant criticism, how ever, is that this research has failed to analyze w hether the re-weighted versions o f the official m onetary aggregates, M l, M2, M3 and L, represent asset groupings that can or should be aggregated. The shaded insert at right illustrates how these issues can affect the behavior o f a m onetary aggregate. 23A Divisia monetary aggregate is constructed in the follow ing manner: Let q„ and plt represent the quantities and user costs of each asset to be included in the aggregate at time t. The expenditure share on the services of monetary asset i in period t is: 8 _ p,q, The user cost of each asset is measured as: P„ = P ;(R ,-r„)(1 -M ) 1 + R,(1 - M) where P*is the geometric mean of the CPI and GNP deflator, R, is the maximum available expected holding period yield, rit is the observed or imputed nominal own rate of return on asset i and M is the average marginal tax rate [see Barnett (1978)]. The growth rate of a Divisia ag gregate then can be written as G(Q,) = j I 1 s* G(qJ, where s* = 1/2(s„ + slit.,), and n is the number of assets in the aggregate. For a more detailed explanation of this weighting procedure and its application to the Divisia ag gregates, see Barnett (1978) and Barnett et al. (1984). 24Barnett (1984), Serletis (1986), Hancock (1987) and Ishida (1984) find the Divisia aggregates to be superior to simplesum measures in a variety of final uses (e.g., stability of velocity, relationship with nominal spending or prices). Bat ten and Thornton (1985), against the standard of perfor mance in nominal spending equations, find little difference between the two. FEDERAL RESERVE BANK OF ST. LOUIS DIVISIA M IA AND THE MONEY METRIC INDEX Previous research that has tested fo r weakly separable asset groups has suggested tw o w eighted m onetary aggregates as possible im provem ents over traditional simple-sum mea sures. These new aggregates are Divisia M IA and the M oney M etric Index (M M I).25 The fo r m er includes only currency and traditional de mand deposit balances, which are w eighted by their user costs to construct a Divisia aggregate.26 The latter includes a w id er range o f assets: cur rency, demand deposits, other (interest-bearing) checkable deposits, savings deposits and o v er night repurchase agreements. The interested reader is directed to the original sources fo r detail on the testing strategies that led to the choices o f these asset groups. The velocities fo r these tw o measures and M l are shown in figure 2. Perhaps the most in teresting aspect o f figure 2 is that, in the early 1980s, w hen M l velocity began its heralded 25The more narrow measure, Divisia M1A, is derived in Belongia and Chalfant (1989). The broader measure, the Money Metric Index (MMI), is discussed in Swofford and Whitney (1990b). Both groupings were determined by nonparametric tests for weak separability described in Varian (1982). While parametric tests for weak separability have been reported by Serletis (1986) and Hancock (1987), Swofford and Whitney (1986) have noted that such tests are sensitive to the choice of the functional form for the utility or production function. 26The user cost formula is derived in Barnett (1978). We use a simplified formula: user cost =-p :(1r ,+ -R.r„) Although the return on something completely illiquid, such as the return on human capital, would be the preferred choice to represent Rt, practical measurement issues have led to the use of long bond rates in empirical studies. The approach used by Swofford and Whitney (1990a) actually calculates upper and lower bounds for the index rather than a unique value for each time period. To avoid any problems associated with choosing an incorrect value of their index, we simply used the Divisia weighting scheme to aggregate the asset collection they identified. 27 Calculating A Divisia Monetary Aggregate Th e example below illustrates the conse quences o f using a m onetary aggregate that is composed o f the "incorrect” assets or that employs simple-sum weighting o f its constitu ent assets even if they are the proper ones to use. For simplicity, w e assume that only four asset categories exist: currency, traditional (non-interest-bearing) demand deposits, NO W accounts (interest-bearing checkable deposits) and savings deposits. W e then define M l as the sum o f C + DD + NOWs and M2 = M l + Savings. From these definitions, w e con struct four aggregates: simple-sum M l and M2 and w eighted versions o f M l and M2. For purposes o f these examples, w e start w ith the follow ing initial conditions: Dollar Balances in Each Asset Category C = 100 DD = 50 NOW s = 50; Savings = 50; M l = 200 M2 = 200 + 50 = 250 Interest Rates on Each Asset Benchmark rate (R) = 0.1; rc = 0; rDD = 0.03; ^now = 0.05, rSAV = 0.05. A simplified expression fo r the user cost formula, cited in footnote 26, is: D _p plt = _ i___ ii. Thus, the user cost fo r each 1 + R, asset is: 1 -0 1 C: — -----— = — = .0909 1 + .1 1.1 1.1 0.0636 NOWs: 1.~ ....... = 0.04545 1.1 SAV: 1 ~ DD: .0636(50) + 14.55 = 21.9 NOWs: .04545(50) + 14.55 = 15.6 Similarly, fo r M2 assets: C: .0909(100) + 16.82 = 54.1 DD: .0636(50) + 16.82 = 18.9 NOWs: .04545(50) + 16.82 = 13.5 SAV: .04545(50) - 16.82 = 13.5 The divisors in each column, 14.55 and 16.82, are the sums o f expenditures on all assets in that group. W ith this information, w e can il lustrate how the tw o issues o f c o n c e r n w eighting and composition—affect the behav ior o f each aggregate. An Illustration: Assume, fo r quarterly data, $10 is shifted from savings to demand de posits. For the simple-sum aggregates, the dollar changes in levels and corresponding grow th rates are: M l: User Costs 1 — ns DD: —----- ^ = C: .0909(100) -*- 14.55 = 62.5 = 0.04545 1.1 100 + (50 + 10) + 50 = 210 Change Grow th rate +10 21.55% M2: 100 + (50+10) + 50 + (50-10) = 250 0 0% Weighted M l : The grow th rate o f a Divisia in dex is constructed as the grow th rate o f each asset category w eighted by the average shares o f the tw o periods.1 N ew budget shares at quarter t + 1 are 59.9 (C), 25.1 (DD) and 15.0 (NOWs). A verage shares (S*) fo r the tw o per iods are 61.2 (C), 23.5 (DD) and 15.3 (NOWs). This yields: 0.612»(0) + 0.235»(grow th rate o f DDs) + 0.153*(0) = 0.235 [(| Q )4 - 1]* 100 Shares Expenditure shares are calculated as s.t = _Pit9h_ SPi.qn xhus, fo r M l assets: = 0.235*(1.0736)* 100 = 25.23% Weighted M2: For this aggregate, the budget share o f DDs rises while that o f savings de i NOVEMBER/DECEMBER 1990 28 clines. Using the average budget shares across the tw o periods and recognizing that the changes in C and NOWs both equal zero, w e have: 0 + .2069 » (107.36) + 0 + .1211* (-59.04) = 22.22 + (-7.15) = 15.07% Thus, a simple $10 shift from one category to another produces growth rates ranging from 'The weighted measures are indexes set equal to 100 in the same base year as opposed to M1 and M2, which are measured in dollars. Thus, comparisons of changes in levels are uninformative. 2ldeally, we would like to isolate the effect of an interest rate change alone, but this is not done for two reasons. zero (simple-sum M2) to 25.23 percent (Divisia M l). The intuition behind these results is that $10 has been shifted from one deposit category (savings) to another with greater "moneyness” because demand deposits have a higher user cost, i.e. their ow n rate o f return is lower. Thus, fo r example, Divisia M2 will show growth while simple-sum M2 will not because $10, while still in M2, has been shifted into a cate gory that receives a larger weight when its growth rate is calculated.2 First, economic theory predicts changes in quantities demanded or supplied in response to changes in relative prices. Second, as can be seen from the calculations for the growth rate of a Divisia index, the index will not change unless quantities change. Figure 2 Logs of Velocity of M1, Divisia M1A, and MMM ’ Velocity data are indexed to a base value of 100 in 1/1963. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 29 Figure 3 Growth Rates of M1, Divisia M1A, and MMI Percent Percent -2 0 1964 66 68 70 72 74 76 78 80 82 84 86 1988 decline, the velocities o f both weighted aggre gates actually increased. Also interesting is the sharp drop in MM I velocity around 1986, at a time w hen M l velocity also fell sharply but Divisia M IA velocity fell only moderately. Since 1986, all three appear to be on similar paths. figure 3 indicates that the behavior o f these ag gregates was much m ore divergent throughout the 1980s than in previous decades. These trends are exhibited in different form in figure 3, which depicts the grow th rate o f each aggregate. Although the three series often track closely, there are notable exceptions. For example, the grow th o f M M I slows sharply in 1978-79 whereas the other measures continue to g ro w at rates near 5 percent. In 1980-81, both w eighted measures show sharply neg ative grow th rates w hile M l grow th continues to be positive; this is particularly interesting because the decline in M M I over the same per iod w ould refute the idea that this episode m erely reflects shifts out o f narrow M IA bal ances into interest-bearing accounts. In general, Long-Run Relationships betw een M onetary Measures and the Price Level Th e empirical perform ance o f alternative m onetary aggregates and their usefulness as targets fo r m onetary policy might be evaluated in a variety o f ways, fo r example, the use o f St. Louis-type spending equations or short-run m oney demand functions w ith tests o f co effi cient stability over time. Either approach is quite sensitive to specification; fo r example, the results appear to depend crucially on both the sample period chosen and the lag lengths used NOVEMBER/DECEMBER 1990 30 fo r all explanatory variables.27 M oney demand equations also are sensitive to the specification o f such things as the scale and interest rate variables and the equations’ short-run dynamics. Such short-run empirical controversies, how ever, are irrelevant fo r our purposes in this study. As Laidler (1990) has pointed out, many key discus sions o f m oney’s effect on economic activity specifically emphasize long-run relationships, testing hypotheses w ith annual or cycle average data, and never argue that economists should be able to m odel the short-run dynamics in herent in quarterly data. T h erefore, w e w ant to focus on one desirable characteristic o f any aggregate monetary vari able: stability in its long-run relationship w ith a variable o f prim ary importance to m onetary policy. If w e return to the equation o f exchange, M V = PT, w e can see that long-run inferences about future movements in the price level (P) can be made from the behavior o f the appropri ate m oney stock (M) if w e have some notion about the long-run behavior o f output (expres sed as the volum e o f transactions, T). For this analytical fra m ew ork to be useful, the velocity o f circulation (V), given some long-run forecast o f output, must be stable. Indeed, this simple fram ew ork fo r predicting the long-run behavior o f inflation is the basis o f the w idely discussed “ P *" analysis o f Hallman, Porter and Small (1989). One w ay to investigate the stability o f velocity is to test w hether its time series has a unit root. I f a time series has a unit root, it is said to be nonstationary; this means that “ shocks” to its path may m ove the series permanently away from its previous pattern. In this case, it may w ander from w hat appeared to be its trend without any force to push it back again. If velocity w e re a time series that did not exhibit some stable long-term behavior, the Quantity Th eory fram ew ork w ould not provide accurate predictions o f long-run inflation on the basis o f m oney stock movements. Velocity, o f course, is calculated as V = (PT)/M. The velocity o f circulation was calculated fo r Divisia M I A and MMI, and, fo r purposes o f comparison, fo r previous (M l) and current (M2) m onetary targets as well. The Augmented Dickey- 27See Friedman and Kuttner (1990) for evidence on the ef fects of a changing sample period and Thornton and Bat ten (1985) for a discussion of the sensitivity of such results to lag length selection. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis Fuller (ADF) test fo r a unit root then was ap plied to each velocity series b y estimating a regression o f the form : (1) AX, = a + bAX,_, + cAX,_2 + dX,., + £„ w h ere X, is the natural logarithm o f the vari able, and A is the difference operator.28 The ADF test examines w hether the coefficient, d, on the lagged level o f X is significantly different from zero. If it is not, then w e cannot reject the null hypothesis that a unit root is present in the data series. Although the estimated t-statistic from this regression can be used in the ADF test, the critical values must be m odified from those o f the standard t-distribution. Because o f the sample sizes in this study and because the null hypothesis is that a unit root is present in the data, the unit root hypothesis w ill be re jected fo r calculated t-values that are negative and exceed 2.90 in absolute value. If the unit root hypothesis is not rejected, equation 1 is re-estimated after including a trend term; in this case the critical value fo r the coefficient d becomes -3 .5 and the null hypoth esis o f a unit root w ill be rejected by a test statistic that is a larger negative number. This test is conducted because a time series may be stationary around a deterministic trend. Finally, because it has been alleged that data from the 1960s dominate relationships betw een m oney and other variables or that data from the 1980s distort the measurement o f M l and M2, the ADF tests w ere conducted over three sample periods: I/1963-IV/1980; I/1970-IV/1989; and I/1963-IV/1989. Table 1A reports the ADF test statistics fo r the logarithms o f each o f the fou r m onetary vel ocities during the three sample periods; the r u and r T columns refer to the cases without and w ith trend, respectively. As the results indicate, no velocity measure is stationary in the log levels even after accounting fo r trend. Thus, w e difference the data again and estimate a regres sion o f the form: (2) A2X, = a + bA2X,_i + cA2X,_2 + dAX,., + e„ w h ere the test again is w hether the coefficient, d, is significantly different from zero. As before, 28For more detail on the test, see, for example, Dickey, et al. (1986), especially pages 16-17. An FPE criterion was used to determine the number of lagged values for AX,_, 31 Table 1A Augmented Dickey-Fuller Tests on Monetary Velocities Monetary variable 1/1963 - IW1980 t u tt M1 M2 Divisia M1A MMI 0.49 -1.25 1.44 1.53 -1.19 -1.99 -0.32 -0.23 1/1970 - IV/1989 T, -1.92 -2.33 -0.64 -0.91 Tt -1.26 -2.33 -1.66 -1.46 1/1963 - IV/1989 T, -1.88 -2.72 -0.34 -0.73 -1.05 -2.69 -1.79 -1.72 Table 1B Augmented Dickey-Fuller Tests on Growth Rates of Monetary Velocities Monetary variable 1/1963 - IV/1980 t u tt 1/1970 - IV/1989 T» Tt M1 -5.84* -5.87* -3.10* -3.29* M2 -4.50* -4.51* -4.34* -4.31* Divisia M1A -5.59* -5.96* -4.55* -4.52* MMI -3.54* -3.86* -3.01* -3.00* NOTE: An asterisk denotes statistical significance at the 5 percent level. if the null hypothesis is not rejected, equation 2 is re-estimated w ith a trend term added. The results o f this estimation are reported in table IB. In this case, the results again find nothing to distinguish any o f the m onetary in dexes as indicators o f long-run trends in infla tion as each measure apparently exhibits a stable grow th rate o f velocity in each sample period. I f one raises the significance level o f the ADF test statistic to the 0.01 level, how ever, critical values fo r the x u and x T tests o f about -3 .6 and -4.15 w ould indicate that only the velocities o f M2 and Divisia M IA show the ex istence o f a single unit root over time. In con trast, doubts w ould be raised about the stability 29As was noted in the paper’s introduction, a monetary ag gregate also must be controllable by the central bank to be useful as an intermediate target variable. Applying a test used by Belongia and Chalfant (1989) to each of the four aggregates used in this article produces the following results: (1) Div M1A, = -2.25 + 1.05 * AMB,; R2 = .34; DW = 1.53 (2.16) (7.49) (2) M1, = -3.21 + 1.37 *AMB,; R2 = .56; DW = 1.39 (3.69) (11.72) (3) M2, = 2.98 + 0.78 * AMB,; R2 = 0.23; DW = 0.86 (2.98) (5.76) (4) MMI, = -4.79 + 1.38 * AMB,; R2 = 0.18; DW = 1.19 (2.32) (4.96) 1/1963 - IV/1989 T„ -3.86* -5.31* -5.38* -3.62* Tt -4.08* -5.29* -5.36* -3.61* o f the grow th rate o f MM I velocity in each o f the three sample periods and about M l velocity at least in the 1970-89 interval; the stability o f M l velocity over the entire 1963-89 sample w ould be a borderline call. The upshot o f these cursory results is that, fo r the specific final use criterion chosen (sta tionary velocity), a badly designed m onetary ag gregate (M2) may perform better than one that is measured in a manner consistent w ith both economic and statistical theory (MMI). Converse ly, some aggregates that have considered the question o f asset collection and weighting (Divi sia M IA ) perform better than some o f the o f ficial aggregates (M l).29 The intuition of this test is to look for a close, contemporaneous relationship between the instrument of monetary control (the monetary base-AMB) and the target. By this standard, Divisia M1A and simple-sum M1 both appear to be controllable. Although the coefficient for AMB is statistically different from zero for M2 and MMI, the fits (R2 = 0.23 and 0.18) are poor and the low DW statistics suggest that other variables may have important influences on this relationship. Overall, these results do not make a strong case for the controllability of either M2 or MMI. NOVEMBER/DECEMBER 1990 32 SUMMARY The breakdow n in the 1980s o f historical rela tionships betw een the grow th rates o f M l and both the price level and nominal income has led researchers to look fo r explanations o f these changes. Some have argued that the historical relationships w ere spurious and that the cur rent experience demonstrates the true relation ship. Others have argued that the traditional relationships still hold after m inor revisions to the estimating equations. Still others have used indirect evidence to make adjustments to the o f ficial measures o f various m onetary aggregates. None o f these efforts has been com pletely suc cessful in resolving what has been term ed “the velocity problem ” . This article has review ed still another ap proach to this problem: constructing m onetary aggregates from first principles, using economic and index num ber theory to determine which assets should be included in an aggregate as w ell as how they should be w eighted in ag gregation. The tw o measures discussed, Divisia M IA and the M oney M etric Index, exhibited quite d ifferen t velocity behaviors; Divisia M IA velocity is consistent w ith a stable long-run rela tionship w ith the aggregate price level w hile M M I velocity appears not to be. Similarly, M2 velocity appears to be stable over time whereas M l velocity does not. If price stability is an im portant objective o f m onetary policy, monitoring the grow th rates o f certain m onetary aggregates can be a viable approach to policymaking. At the same time, there is still no definitive answer in terms o f all its final uses to the question: W hat is money? REFERENCES Afriat, S. N. “The Construction of Utility Functions from Ex penditure Data,” International Economic Review (1967), pp. 67-77. Barnett, William A. “The User Cost of Money,” Economics Letters, Vol. 1 (1978), pp. 145-49. _______ . “ Economic Monetary Aggregates: An Application of Index Number and Aggregation Theory,” Journal of Econometrics (September 1980), pp. 11-48. _______ . Consumer Demand and Labor Supply: Goods, Monetary Assets and Time (Amsterdam: North Holland, 1981). _______ . “ The Optimal Level of Monetary Aggregation,” Journal of Money, Credit and Banking (November 1982, part 2), pp. 687-710. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis _______ . “ New Indices of Money Supply and the Flexible Laurent Demand System,” Journal of Business and Economic Statistics (January 1983), pp. 7-23. _______ . “ Recent Monetary Policy and the Divisia Monetary Aggregates,” The American Statistician (August 1984), pp. 165-72. _______ . “The Microeconomic Theory of Monetary Ag gregation,” in William A. Barnett and Kenneth J. Singleton, eds., New Approaches to Monetary Economics (London: Cambridge University Press, 1987). _______ . “ Developments in Monetary Aggregation Theory,” Journal of Policy Modeling (Summer 1990), pp. 205-58. Barnett, William A., and Seungmook Choi. “A Monte Carlo Study of Tests of Blockwise Weak Separability,” Mimeo, Department of Economics, University of Texas at Austin, December 1986. Barnett, William A., Melvin Hinich and Piyu Yue. “ Monetary Policy with the Exact Rational Expectations Monetary Ag gregates,” in, Michael T. Belongia, ed. Monetary Policy on the 75th Anniversary of the Federal Reserve System (Kluwer Academic Publishers, forthcoming). Barnett, William A., Edward Offenbacher and Paul Spindt. “ New Concepts of Aggregated Money,” Journal of Finance (May 1981), pp. 497-505. _______ . “The New Divisia Monetary Aggregates,” Journal of Political Economy (December 1984), pp. 1049-85. Barnett, William A., and Paul A. Spindt. Divisia Monetary Ag gregates: Compilation, Data and Historical Behavior, Staff Study #116, Board of Governors of the Federal Reserve System, May 1982. Batten, Dallas S., and Daniel L. Thornton. “Are Weighted Monetary Aggregates Better than Simple-Sum M1?” this Review (June/July 1985), pp. 29-40. Belongia, Michael T., and James A. Chalfant. “ The Chang ing Empirical Definition of Money: Some Estimates from a Model of the Demand for Money Substitutes,” Journal of Political Economy (April 1989), pp. 387-97. Clower, Robert W. “ Is There an Optimal Money Supply?” Journal of Finance (May 1970), pp. 425-33. Darby, Michael R., Angelo R. Mascara and Michael L. Marlow. “ The Empirical Reliability of Monetary Aggregates as Indicators: 1983-87,” Economic Inquiry (October 1989), pp. 555-85. Davis, Richard G. “ How Much Does Money Matter? A Look at Some Recent Evidence,” Federal Reserve Bank of New York Monthly Review (June 1969), pp. 119-31. Deaton, Angus, and John Muellbauer. Economics and Con sumer Behavior (London: Cambridge University Press, 1980). de Leeuw, Frank, and John Kalchbrenner. “ Monetary and Fiscal Actions: A Test of Their Relative Importance in Economic Stabilization—Comment,” this Review (April 1969), pp. 6-11. Dickey, David A., William R. Bell and Robert B. Miller. “ Unit Roots in Time Series Models: Tests and Implications,” The American Statistician (February 1986), pp. 12-26. Ewis, Nabil A., and Douglas Fisher. “ The Translog Utility Function and the Demand for Money in the United States,” Journal of Money, Credit, and Banking (February 1984), pp. 34-52. 33 _______ . “Toward a Consistent Estimate of the Substitut ability Between Money and Near Monies,” Journal of Macroeconomics (Spring 1985), pp. 151-74. Fayyad, Salam K. “A Microeconomic System-Wide Approach to the Estimation of the Demand for Money,” this Review, (August/September 1986), pp. 22-33. Feenstra, Robert C. “ Functional Equivalence Between Li quidity Costs and the Utility of Money,” Journal of Monetary Economics (March 1986), pp. 271-91. Fisher, Irving. The Purchasing Power of Money, Reprints of Economic Classics (Augustus M. Kelley, 1963). Friedman, Benjamin M. “ Even the St. Louis Model Now Believes in Fiscal Policy,” Journal of Money, Credit, and Banking (May 1977), pp. 365-67. Friedman, Benjamin M., and Kenneth N. Kuttner. “ Money, Income, Prices and Interest Rates After the 1980s,” Work ing Paper, WP-90-11 Federal Reserve Bank of Chicago (July, 1990). Friedman, Milton, and Anna J. Schwartz. “ Money and Business Cycles,” The Review of Economics and Statistics (February 1963, Part 2), pp. 32-64. _______ . Monetary Statistics of the United States (Columbia University Press, 1970). Garfinkel, Michelle R. “The FOMC in 1989: Walking a Tightrope,” this Review (March/April 1990), pp. 17-35. Goldfeld, Stephen M., and Alan S. Blinder. “Some Implica tions of Endogenous Stabilization Policy,” Brookings Papers on Economic Activity (3:1972), pp. 585-640. Gordon, Robert J. “The Short-Run Demand for Money: A Re consideration,” Journal of Money, Credit and Banking (Part 1, November 1984), pp. 403-34. Gramlich, Edward M. “The Usefulness of Monetary and Fiscal Policy as Discretionary Stabilization Tools,” Journal of Money, Credit, and Banking (May 1971), pp. 506-32. Hafer, R. W. “The Money-GNP Link: Assessing Alternative Transactions Measures,” this Review (March 1984), pp. 19-27. Hafer, R. W., and Joseph H. Haslag. “The FOMC in 1987: The Effects of a Falling Dollar and the Stock Market Col lapse,” this Review (March/April 1988), pp. 3-16. Hallman, Jeffrey J., Richard D. Porter and David H. Small. “ M2 Per Unit of Potential GNP as an Anchor for the Price Level,” Staff Studies, No. 157, Board of Governors of the Federal Reserve System, April 1989. Hancock, Diana. “Aggregation of Monetary Goods: A Pro duction Model,” in William A. Barnett and Kenneth J. Singleton, eds., New Approaches to Monetary Economics (London: Cambridge University Press, 1987). Hetzel, Robert L. “ M2 and Monetary Policy,” Federal Reserve Bank of Richmond Economic Review (September/October 1988), pp. 14-29. Ishida, Kazuhiko. “ Divisia Monetary Aggregates and Demand for Money: A Japanese Case," Bank of Japan Monetary and Economic Studies (June 1984), pp. 49-80. Klein, Benjamin. “ Competitive Interest Payments on Bank Deposits and the Long-Run Demand for Money,” American Economic Review (December 1974), pp. 931-49. Laidler, David. “The Legacy of the Monetarist Controversy,” this Review (March/April 1990), pp. 49-64. _______ . “ The Quantity Theory is Always and Everywhere Controversial—Why?,” Seminar Paper 89-03, Centre for In ternational Economic Studies, University of Adelaide (July 1989). Offenbacher, Edward K. The Substitutability of Monetary Assets, unpublished Ph.D. thesis, University of Chicago, 1979. _______ . “ Disaggregated Monetary Asset Demand Systems: Estimation and an Application of the Preference Independence Transformation,” Special Studies Paper No. 144, Board of Governors of the Federal Reserve System, July 1980. Serletis, Apostolos. “ Monetary Asset Separability Tests,” in William A. Barnett and Kenneth J. Singleton, eds., New Ap proaches to Monetary Economics, (London: Cambridge University Press, 1987). Serletis, Apostolos, and A. Leslie Robb. “ Divisia Aggregation and Substitutability Among Monetary Assets,” Journal of Money, Credit, and Banking (November 1986), pp. 430-46. Spindt, Paul A. “ Money is What Money Does: Monetary Ag gregation and The Equation of Exchange,” Journal of Political Economy (February 1985), pp. 175-204. Stone, Courtenay C., and Daniel L. Thornton. “ Solving the 1980s’ Velocity Puzzle: A Progress Report,” this Review (August/September 1987), pp. 5-23. Swofford, James L., and Gerald A. Whitney. “ Flexible Func tional Forms and the Utility Approach to the Demand for Money: A Nonparametric Analysis,” Journal of Money, Credit and Banking (August 1986), pp. 383-89. _______ . “ Nonparametric Tests of Utility Maximization and Weak Separability for Consumption, Leisure, and Money,” Review of Economics and Statistics, (August 1987), pp. 458-64. _______ . “A Comparison of Nonparametric Tests of Weak Separability for Annual and Quarterly Data on Consump tion, Leisure and Money,” Journal of Business and Economic Statistics (April 1988), pp. 241-46. _______. “ Bounding an Economic Monetary Aggregate Under Nonhomothetic Preferences,” Journal of Business and Economic Statistics (January 1990a), pp. 137-42. _______ . “A Money Metric Index of an Economic Monetary Aggregate,” Economic Inquiry (1990b), forthcoming. Tatom, John A. “Are the Macroeconomic Effects of Oil-Price Changes Symmetric?, “ in Karl Brunner and Allan H. Meltzer, eds., Stabilization Policies and Labor Markets, Carnegie-Rochester Conference Series on Public Policy, (Spring 1988), pp. 325-68. Thornton, Daniel L., and Dallas S. Batten. “ Lag-Length Selection and Tests of Granger Causality Between Money and Income,” Journal of Money, Credit, and Banking (May 1985), pp. 164-78. Varian, Hal R. “The Nonparametric Approach to Demand Analysis,” Econometrica (July 1982), pp. 945-74. _______. “ Non-parametric Tests of Consumer Behaviour,” Review of Economic Studies (January 1983), pp. 99-110. Wenninger, John. “The M1-GNP Relationship: A Component Approach,” Federal Reserve Bank of New York Quarterly Review (Autumn 1984), pp. 6-15. NOVEMBER/DECEMBER 1990 34 Alison Butler Alison Butler is an economist at the Federal Reserve Bank of St. Louis. Lora Holman provided research assistance. The Trade-Related Aspects of Intellectual Property Rights: What Is At Stake? The pirating o f U.S.-financed research and developm ent discourages innovation, denies markets to Am erican exports, and threatens technological progress. Protection o f intellectual property rights preserves Am erica’s technological edge, which is a key to our continued international competitiveness.” —Clayton Yuetter, U.S. trade representative, 1986 J L HE SENTIMENTS expressed by the form er U.S. trade representative above could have easily been made by trade representatives from any industrialized country about their ow n country. In the last fe w years, many industrial countries have becom e increasingly concerned over the lack o f international protection o f intellectual property rights (IPR). In a 1988 study, the U.S. International Trade Commission (USITC) attempted to estimate the econom ic effects o f intellectual property rights infringem ent by foreign countries on U.S. firm s.1 In its survey, firm s estimated their losses at $6.2 billion in exports sales in 1986. These losses accounted fo r approximately 1.4 percent o f the total export sales o f products which are covered by some form o f intellectual property rights that year. Firms estimated additional losses o f $1.8 billion in sales in 1986 due to imports that in 1USITC (1988). For limitations of the survey and data, see the study. 2See also Feinberg (1988). FEDERAL RESERVE BANK OF ST. LOUIS fringed on their domestic IPR. In a related study, Feinberg and Rousslang (1990) estimate that the ratio o f lost profits to w orld w id e sales o f firms selected from the USITC study ranged from 0.05 percent fo r the extractive sector to 3.6 per cent fo r the entertainment sector.2 Since firms incur research and developm ent (R&D) expenses to develop intellectual property, these lost sales and profits defer firm s from undertaking the risky process o f developing intellectual property in the future. Frustrations over the lack o f effective interna tional agreements regarding IPR have led the United States to unilaterally create measures to deal w ith w hat it perceives as unfair trading practices. In particular, the so-called “ Special 301” clause o f the Omnibus Trade and Com petitiveness Act o f 1988 requires the U.S. Trade Representative to identify countries that do not 35 adequately protect or enforce IPR. The trade representative then has the authority to bring an unfair trade practice case against that country. Although several countries have been put on a “priority watch list,” no cases have yet been in itiated by the U.S. Trade Representative under the “ Special 301” designation. Countries throughout the w orld, particularly the m ore industrialized ones, are interested in increasing the amount o f protection and en force ment o f IPR internationally. Th eir concerns are seen in the current negotiations o f the General Agreem ent on Tariffs and Trade (GATT). G A TT is the principal rule-making body fo r interna tional trade, whose goal is to eliminate trade barriers that reduce the fre e flo w o f goods. In the current round o f negotiations, the so-called Uruguay Round scheduled to end in Decem ber 1990, one o f the 15 negotiating groups is devoted to developing an agreem ent regarding "traderelated aspects o f intellectual property rights, including trade in counterfeit goods.”3 Current ly, intellectual property rights are explicitly ex cluded from G A T T ’s auspices. There are many economic and legal issues related to the protection o f intellectual prop er ty.4 The analysis in this paper focuses only on the trade-related aspects o f IPR, that is, the e f fect o f differential (and uncertain) IPR across countries on trade and the benefits and costs o f creating international standards fo r protecting intellectual property. INTELLECTUAL PROPERTY Intellectual property is an invention, idea, p ro duct or process that has been registered with 3GATT (1990), p. 11. “For a discussion of some of the legal issues regarding IPR in international trade, see Meessen (1987). To narrow the focus of the paper, two issues have been ignored: which items should be protected by intellectual property regula tions and what type of protection—patents or copyrights— is appropriate for certain types of goods, such as software. HJSITC (1988). 6Other types of intellectual property, such as trade secrets and mask works, constitute a small percentage of all in tellectual property and are not discussed here. In addition, because there are no internationally agreed-upon defini tions for intellectual or industrial property, these definitions should be considered chiefly as guidelines. intellectual property protection is important insofar as in novation is desirable. Some analysts argue (see, for exam ple, Nordhaus, 1969; Grossman and Helpman, 1989) that innovation increases growth worldwide, improving the the governm ent and that awards the inventor (or applicant) exclusive rights to use the inven tion fo r a given period o f time. It confers “ ...the right to exclude others from making, us ing, or selling the invention w ithin the national territory."5 (For a discussion o f w hy intellectual property is awarded protection, see shaded in sert on pages 39 and 40.) Industrial property protection is generally awarded only to new and useful products and production processes. Intellectual property is protected by the govern ment in a variety o f ways. Copyrights are awarded to protect original works o f author ship, such as literary, artistic and musical works; trademarks allow a manufacturer ex clusive rights to a distinguished name, symbol or mark.6 As noted above, these rights are only valid in the countries in which a patent applica tion has been awarded. This lack o f an interna tional IPR system could have a significant im pact on the amount o f innovative activity.7 CURRENT PROTECTION OF IPR The protection awarded different types o f in tellectual property varies substantially across countries. As shown in table 1, most countries have patent lengths o f 15 years or m ore.8 His torically, an innovation was awarded patent pro tection fo r the num ber o f years approximately equal to the amount o f time it took to train tw o apprentices (14 years).9 Although the apprentice system has been obsolete fo r many years, patent lengths have rem ained essentially unchanged. Often, how ever, the actual period in which a firm can sell its product under patent protection is shortened considerably; fo r example, tests fo r quality of life through such things as better medicines, im proved living conditions and safer production processes. Of course all innovations may not be beneficial, nor do all people necessarily benefit from all innovations (see, for ex ample, Kamien and Schwartz, 1982 and Maskus, August 1990). In this paper, it is assumed that innovation has a net positive effect on a country’s growth and on growth worldwide. 8AII countries referred to are members of the World Intellec tual Property Organization (WIPO). All statistics in this sec tion, unless otherwise cited, are from WIPO (1988). 9Benko (1987). NOVEMBER/DECEMBER 1990 36 product safety (which, fo r pharmaceuticals, can take up to 10 years) are included in the life o f a patent. Recently, both the United States and the European Community proposed legislation that w ould increase the effective length o f patent protection—that is, the amount o f time a firm can actually market or use a product before its protection expires.10 Table 1 Domestic Protection Number of countries General length Patents1 18 28 57 5-10 years 10-15 years 15-20 years Copyrights2 8 75 9 25 years 50 years 50+ years Intellectual property Several additional factors affect the amount o f protection intellectual property is aw arded in any given country. As tables 2 and 3 show, many countries, including some industrial ones, ex clude certain industries from patent protection. Those excluded are prim arily high-technology industries w ith very high R&D intensities that tend to produce expensive products. Although currently a substantial amount o f innovation oc- 'From date filed. 2From author’s death. SOURCE: World Intellectual Property Organization (1988) 10See Congress and the Nation (1985) and “ Patents for Pharmaceuticals” (1990). Table 2 Patent Exclusions_______________ Product Pharmaceutical products (PHARM) Animal varieties (ANIM) Methods for treatment of humans or animals (TREAT) Plant varieties (PLANT) Biological processes producing plant or animal varieties (BIO) Food products (FOOD) Computer programs (COMP) Chemical products (CHEM) Nuclear inventions (NUC) Pharmaceutical processes Food processes Microorganisms Substances from microbiological processes Cosmetics Fertilizers Mixture of metals and alloys Agricultural machines Anticontaminants Methods of agriculture or horticulture Number of countries that exclude Industrial countries1 47 59 8 18 17.0% 30.5 59 57 18 18 30.5 31.6 56 34 48 21 13 18 8 20 2 2 32.1 23,5 41.7 9.5 15.4 10 9 8 2 3 1 20.0 33.3 12.5 6 2 2 2 1 1 1 0 0 0 0 0 16.7 0.0 0.0 0.0 0.0 0.0 1 0 0.0 Percent industrialized 1These are the Western European countries, Japan, Canada, Australia, United States, New Zealand and Iceland. SOURCE: World Intellectual Property Organization (1988) http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 37 Table 3 Selected Patent Exclusions by Country Industrialized Countries Australia Austria-Belgium Canada Denmark Finland-Norway France Germany Greece Iceland Ireland Italy Japan Luxembourg Netherlands New Zealand Portugal Spain Sweden Switzerland2 United Kingdom United States Total PHARM ANIM TREAT PLANT X X X X X X X1 X X X X X X X1 X X X X X X X’ BIO X X X X FOOD X X X X X X X X’ X X X COMP CHEM NUC X X X X X X X X1 X X X X X’ X X' X X X X X X X X X X X X X X X X X X X X X X X X 8 18 18 18 18 8 20 2 2 13 6 8 6 5 7 2 6 3 X X X X X X X X X X X 8 20 18 20 20 5 18 1 0 Western Hemisphere Brazil Mexico Venezuela 10 8 X X 7 X X 8 8 X X X X X 8 2 X X 6 X X X 3 European Eastern 8 7 7 4 8 5 5 4 5 2 6 6 6 3 6 6 5 5 Total 39 41 41 39 38 26 28 19 11 TOTAL 47 59 59 57 56 34 48 21 13 X X X X X X X X' X1 X X X X X X3 X X X X X Developing Countries Asian India Indonesia Pakistan Rep. of Korea Thailand African X X X X X X X X X X X X X NOTE: See table 2 for definitions. 1lf filed through the European Patent Convention (EPC). includes Liechtenstein Liechtenstein excludes if filed through the EPC. SOURCE: World Intellectual Property Organization (1988) NOVEMBER/DECEMBER 1990 38 curs in some excluded industries—fo r example, genetic engineering—it generally takes place in countries that do not exclude them from protec tion. For example, although European firm s ac count fo r 82 percent o f w orld investment in in dustrial plant and other biotech assets, only 2 percent o f it was spent in Europe. Similarly, 77 percent o f all patents in biotechnology w ere issued in the United States and Japan, which o f fe r the most extensive patent protection.11 The length o f copyright protection, on the other hand, is a m ore standardized measure across countries, w ith the m ajority o f countries issuing copyright protection fo r the rem ainder o f the author’s life plus 50 years. Other factors that influence the extent o f pro tection fo r intellectual property are the en force ment o f existing laws and the amount o f copy right and patent protection in other countries. These factors are particularly important w hen examining the trade-related aspects o f IPR. Company $5.7 billion; this amount was reduced to $909.5 million in 1990 on appeal. Th e second type o f patent infringem ent w ould occur if a company produced a camera similar to Polaroid's and labeled and sold it as a Polaroid. This is an example o f a "cou nterfeit” product. In this instance, Polaroid’s tradem ark as w ell as its patents w ould be infringed upon. International trade can affect even non-traded products that are protected by IPR. For example, suppose a pharmaceutical firm , call it SAW, re cently developed a new product called NOCOLD, that cured the common cold and had no side e f fects. If the firm intended only to sell NOCOLD in its ow n country (call it the North), it w ould file fo r a patent only in the North. The problem w ith the lack o f an international system o f rules regarding IPR occurs w hen the cost o f copying an innovation (including the cost o f penalties if caught) is less than that o f either purchasing or leasing the technology itself. Even if SAW does not export NOCOLD, h ow ever, the rate o f return earned could be affected b y international trade. For example, a firm in another country could create either a counter feit version or a cheaper imitation o f the p ro duct and export it to the North.12 Similarly, a country that is producing NOCOLD legally could export it back to the North.13 I f sales o f the foreign-produced product in the North reduce SAW ’s sales o f NOCOLD and its profits on the product, the firm w ould expect a low er return on NOCOLD as w ell as all prospective products. Because the expected rate o f return affects the decision to develop new products, the firm undertakes fe w e r projects in the future. Thus, protecting intellectual property is not just a domestic issue. Th ere are prim arily tw o ways that a product patent can be infringed upon: First, a copy o f the product can be produced, w ith essentially the same characteristics (although not necessari ly the same quality) as the original, but w ith no pretense o f being the original product. Typical ly, such copies are sold at a low er price. An ex ample o f this occurred in 1976 w hen Kodak in troduced a line o f instant cameras that w ere similar to those already patented by Polaroid. In 1986, U.S. courts ruled that Kodak had infringed on Polaroid’s patents and awarded the Polaroid I f the company that infringed upon a patent in either o f the tw o ways was not a U.S. com pany, the patent laws in both the United States and the country in w hich it was located w ould have to be considered. Unless an agreement states otherwise, a patent or copyright is valid only in the country in which it was issued.14 Thus, if Kodak had produced and sold its cam eras in countries that did not honor Polaroid’s patent, Polaroid w ould have been unable to sue Kodak. Because many patent violations oc cur across national borders, differences in pa- TRADE-RELATED ASPECTS OF INTELLECTUAL PROPERTY: SOME THEORETICAL ISSUES Why IP R Is an International Issue 11“ Bugs that Divide” (1990). >2Not all countries have made it illegal to import products that infringe on domestic patents. 13Parallel trade, which occurs when a product produced under a patent in one country is exported to another coun try and competes with the patented product in that coun try, is legal in many countries and for intra-EC trade. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis 14There are surprisingly few agreements among countries to honor each other's patents. This is discussed in greater detail later. 39 The Economics of Innovation Th ere are essentially tw o types o f techno logical innovations: process innovations, which are new production processes or im prove ments on existing technology, and product in novations, which are the creation o f new p ro ducts or improvements on existing products.1 Both types o f innovation are patentable. Be cause the economics o f these tw o are essen tially the same, the discussion focuses on p ro duction innovation fo r simplicity. Intellectual property has the unusual (al though not unique) property that the know ledge it contains is not depleted w ith use. For example, no matter how many times the fo r mula fo r aspirin is used, the formula itself (that is, the knowledge contained in the pa tent) remains unchanged. As a result, the marginal cost o f using this knowledge (e.g., the formula fo r aspirin) is zero. For economic efficiency, this knowledge should be made available to anyone interested, because doing so does not diminish the stock o f knowledge (or reduce the number o f times aspirin can be made). O ver time, how ever, such a policy would have some unfortunate consequences. Generally, innovation is the result o f invest ment expenditures on research and develop ment (R&D). Because expenditures on R&D occur before a new product is created, the firm's decision to incur these costs involves considerable uncertainty. The expected rate o f return on R&D, which is the present dis counted value o f the stream o f net operating profits divided by the present value o f the R&D costs, has to be at least as great as the opportunity cost o f resources devoted to R&D—that is, the expected rate o f return that would have been earned if the same resources allocated to R&D w e re invested elsewhere. W hile the opportunity cost o f capital is easy to determine (it is simply the interest rate), the rate o f return on R&D is m ore difficult to ascertain. It depends on h ow much R&D must be spent before a new product is dis covered and developed, how much demand there w ill be fo r the new product, and how much production costs w ill be. The return on R&D also depends on the time the firm can produce the product exclusively and there fo re earn econom ic profits. W hile the first three factors are outside the scope o f this paper, the last factor demonstrates how IPR can effect the incentive to innovate. In the absence o f governm ent intervention, maintaining exclusive rights to an innovation fo r any period o f time is often difficult. Given that the marginal cost o f using the knowledge created b y the innovation is zero, one could conclude that governm ents have no reason to award these rights. W ithout assigning ex clusive rights to produce the innovation, h ow ever, the amount o f time the innovating firm can produce the product is both less certain and likely shorter; any other firm that can figure out how to make the product could also produce it without changing the know ledge associated with the innovation. For ex ample, a firm that did not discover the fo r mula fo r aspirin but, instead, was able to produce it w ould reduce the return earned by the innovating firm. This is true even though entry by the noninnovating firm in this market does not diminish the innovating firm ’s ability to produce aspirin. This reduced return on the investment in R&D appears to increase efficiency by prom oting competition; how ever, it also reduces the number o f R&D projects that w ill be undertaken. If, however, the governm ent assigns property rights to 1The economic reasons for copyright protection, essen tially the same as those for patent protection, are not discussed separately. The reasons for protecting trademarks, however, differ from those for patent and copyrights, as trademarks are thought to provide infor mation for consumers about the quality of a product. Protection of trademarks is intended to ensure that they have some informative value to consumers, rather than to protect an idea itself. NOVEMBER/DECEMBER 1990 40 innovations (and enforces them), then the amount o f time the product can be produced exclusively w ill increase, raising the rate o f return on R&D, which in turn has a positive effect on the amount o f innovation.2 O f course, the w orld is not certain. Th ere is no w ay o f knowing in advance w hether the R&D expenditures w ill produce an eco nomically viable product. Intellectual prop er 2There is some disagreement as to how much IPR in creases innovation (see, for example, Maskus, 1990). The question is a difficult one to measure empirically, because it requires determining what a firm would do in a situation that has not occurred. One attempt to do so is Mansfield (1986), who finds some role for patents in encouraging innovation. tent laws across countries and the lack o f an in ternational enforcem ent system affects the in centives associated w ith innovation. IP R and International Trade: An Example Suppose that SAW wanted to sell NOCOLD in ternationally as well. A firm can do this one o f three ways, all o f which are affected by the state o f international property rights. For simplicity, the example below compares the case in which only one o f the tw o countries protects IPR. D ir e c t E x p o r t s —First, SAW can export NOCOLD directly to another country, which w e w ill call the South. If the product is not p ro tected by a patent in the South, either because SAW has not filed fo r protection there or because the South does not protect IPR, cheaper copies that use the same formula as NOCOLD could be legally sold. Similarly, if the trademark is not protected, counterfeit products that are indistinguishable from NOCOLD can also be legally sold. These copies decrease SAW's total sales and profits associated w ith producing and selling NOCOLD. In addition, if the product has a tradem ark that is strongly associated with SAW, the counterfeit product, if o f significantly poorer quality, could adversely affect the reputation o f the firm and further harm the sales o f both NOCOLD and future products o f SAW. 15For a discussion of the effects of international piracy in motion pictures in the United States, see USITC (1988) and Plock (1989). FEDERAL RESERVE BANK OF ST. LOUIS ty rights are a w ay o f rew arding firms fo r in curring the risk associated w ith R&D by in creasing the expected rate o f return on R&D, thereby making m ore projects possible. As long as innovation is considered desirable, assigning property rights to in tellectual property is one w ay to encourage firms to innovate.3 3For a general discussion of the role of R&D on innova tion, see Kamien and Schwartz (1982). In addition, the issue of determining the optimal length and breadth of a patent (within a country), has recently been dis cussed in the RAND Journal of Economics (Spring 1990) and Economic Inquiry (October 1984). L ic e n s in g —SAW can also license the technolo gy to produce NOCOLD to a firm in the South. Th e Southern firm, in turn, pays SAW a royalty fee. W ithout protection o f IPR, how ever, a firm has no incentive to pay licensing fees if it is cheaper to copy the product than pay those fees. For industries w h ere licensing represents a significant proportion o f its revenues, the lack o f IPR can cause a substantial loss o f revenue fo r a firm . For example, motion picture firms receive considerable revenues from licensing fees paid by foreign distributors. Every “pirated” copy o f a m ovie that is sold or shown represents a loss o f revenue fo r the m ovie com pany.15 F o r e ig n D ir e c t In v e s tm e n t—The third possi bility, called foreign direct investment, occurs if SAW locates production facilities in the South to produce NOCOLD. Again, how ever, the lack o f IPR in the South has a similar effect on innova tion as discussed previously. Because SAW has no guarantee that it can control the production o f NOCOLD in that country, the expected rate o f return on the innovation is smaller, reducing the profitability o f foreign investment in the South; as a result, fe w e r firm s w ill engage in such foreign direct investment. A country’s lack o f adequate protection o f IPR could be particu larly costly in this case because the country is foregoing some o f the benefits o f foreign direct investment, such as n ew resources, training and 41 employment. Obviously, counterfeiting firms might provide some o f these same benefits; since they develop no new products themselves, how ever, they are dependent on others fo r in novations to copy.16 As long as the direct cost o f counterfeiting (or copying), including the likelihood and penalties associated w ith being caught, is less than the profits earned by the firm s doing the copying, firm s w ill continue to pirate technology. Copy ing, how ever, low ers the rate o f return earned by innovators and th erefore the overall amount o f innovation. In the long run, the resulting decline in the amount o f innovation reduces the counterfeiting that can be done by firms in the South. On an aggregate level, the citizens o f both countries end up w orse off. Th ere is less innovation, fe w e r products overall, and w o rld w ide grow th is th erefore lower. W hile the vast majority o f countries have laws protecting intellectual property, the arguments presented above describe what occurs in coun tries that either have w eak IPR or simply fail to enforce w hatever IPR they do have. Given the apparent advantages o f protecting intellectual property, w h y are there so many problems with trade in goods affected by intellectual property rights? Part o f the answer to this question can be found by examining the current international system o f regulating intellectual property rights. CURRENT SYSTEM OF INTERNA TIONAL IPR The main organization responsible fo r interna tional agreements on IPR is the W orld Intellec tual Property Organization (WIPO), which is ad ministered under the auspices o f the United Na tions. W IPO's objectives include administering international treaties and agreements on intellec tual property rights and encouraging the protec tion o f IPR w orld w id e.17 Currently, 125 coun tries are members o f W IPO .18 Table 4 provides a partial list o f these countries and o f signatories 16Whether or not foreign direct investment is desirable for developing countries is a separate issue and is discussed elsewhere. (See, for example, Hood and Young, 1979.) For the purposes of this paper, we assume that the lack of IPR does not mean a country doesn't want foreign direct investment—there are more straight-forward ways to reduce or eliminate foreign investment. 17For a list of organizations that are solely administered by o f several treaties administered by W IPO. The tw o major international agreements on IPR are the Paris and Berne Unions, w hich deal w ith in dustrial property (including patents) and copy rights, respectively. (For a description o f the tw o agreements, see shaded insert on page 43.) In addition to these unions, other international property right agreements administered by W IPO cover such things as industrial design, satellite transmissions and registration o f trade and ser vice marks.19 Not all members o f W IPO, h o w ever, are signatories to these unions and treaties, nor are the number o f signatories fo r each agreement identical (see table 4).20 Differing Patent Regulations One important issue regarding the trade-related aspects o f intellectual property is the problem o f differential patent regulations across coun tries. For example, w hile most countries deter mine patent eligibility based on a first-to-file basis, the United States employs a first-to-invent rule. As a result, patent protection fo r the same invention could be awarded patent protection to different applicants depending on w hether the actual inventor was the first to file. W IPO has had some success in trying to stan dardize the process o f obtaining patents interna tionally through the Patent Cooperation Treaty (PCT). This treaty allows applicants to file fo r a patent in a central office and specify in which o f the signatory countries (shown in table 4) it wishes the application to have effect. This p ro cess reduces the costs o f filing fo r patents by centralizing the search and examination w ork associated w ith determining patent eligibility. The PCT is also designed to increase the time an applicant has to decide w hether to w ithdraw the application fo r foreign patents. A firm might choose to w ithdraw its patent application and avoid the expense associated w ith translating the patent application into the local language and finding a local patent agent if the demand fo r the product to be patented is likely to be too small. 18These statistics exclude the German Democratic Republic because of the unification with the Federal Democratic Republic that occurred on October 3, 1990. 19See WIPO (1990a). 20The United States, for example, did not sign the Berne Union until 1988. I, see W lr U (1988), p. NOVEMBER/DECEMBER 1990 42 Table 4 Signatures of Selected International Intellectual Property Agreements Industrialized Countries Australia Austria-Belgium Canada Denmark Finland-Norway France Germany Greece Iceland Ireland Italy Japan Luxembourg Netherlands New Zealand Portugal Spain Sweden Switzerland2 United Kingdom United States Total WIPO X X X X X X X X X X X X X X X X X X X X X Members of Treaties Berne Paris1 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X PCT X X X X X X X X X X X X X X X X 24 24 24 19 Asian3 India Indonesia Pakistan Rep. of Korea Thailand 27 9 17 4 African 40 25 35 14 Western Hemisphere Brazil Mexico Venezuela 23 14 11 2 European Eastern 11 7 11 7 12 7 4 3 Total 101 59 75 21 TOTAL 125 83 99 43 Developing Countries X X X X X X X X X X X X X X X X X X 'Dominican Republic, Iran, Nigeria, San Marino and Syria are not members of WIPO. includes Liechtenstein. 3Hong Kong, Singapore and Taiwan are not members of WIPO. SOURCE: World Intellectual Property Organization (1990a) http://fraser.stlouisfed.org/ FEDERAL RESERVE Federal Reserve Bank of St. Louis BANK OF ST. LOUIS X 43 The Paris and Berne Unions: Some Background Inform ation “Th e Paris Union, the official name o f w hich is the International Union f o r the Protection o f Industrial Property, was founded b y a Conven tion signed in Paris in 1883 and last revised in 1967. Under the Convention, each m em ber State must accord the same protection to the in ventions, trademarks and other subject mat ter o f industrial property o f the nationals o f the other m em ber States as it accords to those o f its ow n nationals. The Convention also provides certain facilities to foreigners; fo r example, it allows them, without losing their claim to novelty, to file their applica tions fo r patents up to a year after first filing in the country o f origin. It contains provi sions concerning the conditions under which a State may license the use o f a patented in vention in its ow n territory if, fo r example, Ninety percent o f all patent applications are filed in the 43 countries that make up the m em bership o f the PCT.21 The usefulness o f the PCT is demonstrated by the surge in the number o f applications received by the PCT, from 2,625 in 1979 to 14,874 in 1989.22 In addition, the number o f countries designated fo r patent p ro tection by applicants rose from 6.6 percent to 15.8 percent o f m em ber countries during the same time period. Enforcem ent o f IP R As mentioned previously, the enforcem ent o f existing IPR is another big problem. Although W IPO administers several international accords regarding IPR, fe w laws currently en force these treaties. The Paris Union only requires signato ries to give foreigners national treatment, that 21The statistics in this section are from WIPO (1990a, 1990b). 22The potential of this agreement has just begun to be realized. For example, the United States, which is a signatory of the PCT, had 66,850 applications for patents by foreigners. (U.S. Department of Commerce, Patent and Trademark Office, 1989). the ow ner o f the patented invention does not exploit it in such territory. Th e Berne Union, the official name o f which is the International Union f o r the Protection o f Literary and A rtistic Works, was founded by a Convention signed in Berne in 1886 and last revised in 1971. Under the Convention, each m em ber State must accord the same protection to the copy right o f the nationals o f the other m em ber States as it accords to that o f its ow n nation als. Th e Convention also prescribes some minimum standards o f protection; fo r exam ple, that copyright protection generally con tinues throughout the author's life and fo r 50 years thereafter. It includes special provisions fo r the benefit o f developing countries.” W IPO, General Information, 1990 is, award them the same rights as they give their ow n citizens. The Berne Convention includes tw o en force ment provisions. First, it established the p re sumption o f authorship so that the author does not have to prove it; rather, challengers o f the copyright w ould have to provide evidence to the contrary. Second, copies that infringe on a copyright are subject to seizure in any country in which the w ork has a copyright. This also applies to imports o f reproduced materials from countries w h ere the w ork is not protected. W IPO does not have an international dispute settlement mechanism w hereby an applicant (or country) can file a complaint against another country’s implementation o f the treaties. The only recourse under the agreem ent is to bring a case before the International Court o f Justice.23 That court, how ever, can only arbitrate cases 23The court’s ruling is not binding for all members, however. See WIPO (1988). NOVEMBER/DECEMBER 1990 44 that relate to the interpretation or application o f the Paris and Berne Unions. Th ere appears to be no penalty fo r ignoring the ruling o f the Court. W IPO does require countries to give applicants access to the same legal remedies fo r patent in fringem ent as they do their ow n nationals. This requirement, how ever, is subject to existing laws fo r patent violations. In fact, many coun tries do not have explicit penalties associated w ith violations o f IPR, and fe w impose civil penalties. According to W IPO, as o f 1988, only 14 countries had laws requiring the seizure o f infringing patent articles, 20 countries granted compensation o f damages and 11 destroyed in fringing goods. Th ere is somewhat m ore protec tion fo r products w ith trademarks. W IPO is now preparing a study that w ill examine the possibility o f establishing a new treaty which creates a dispute mechanism to arbitrate possi ble violations o f international IPR agreements. Other Issues The problems described above explain w h y there is considerable concern about the current international system o f IPR. Having an interna tional set o f rules regarding the protection o f intellectual property reduces the uncertainty associated w ith innovation and increases the ex pected return earned on those innovations. This does not mean that all countries must have the same degree o f protection—GATT, fo r example, has different rules regarding the acceptable amount o f tariffs fo r industrial and developing countries. Rather, an explicit set o f agreements, along with an effective mechanism to mediate disputes, could significantly decrease the loss o f earnings associated w ith copying and counter feiting innovated products. In addition, such agreements could reduce the information costs o f determining the amount o f protection o f in tellectual property. These costs can be substan tial fo r firm s that intend to sell their products internationally, and, as described above, must file fo r a patent in each country w h ere it wants to sell the product. Similarly, given the current system in which countries can choose to ex 24This terminology comes from Maskus (1990). This analysis can also be used within a country, where some regions have industries with a lot of innovation (such as the Silicon Valley in California), and others have very few innovating industries. For an example of a model that looks at innova tion and technology transfer both within and across coun tries, see Butler (1990). 25Chin and Grossman (1988), for example, find that the http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis clude specific products from patent protection, firm s are faced w ith the choice o f selling their products in some markets w ith no protection or avoiding certain markets altogether. Developing countries that protect IPR w ill increase its access to new technology because innovating firms w ill have stronger incentives to produce and sell their products in countries that protect intellec tual property. PROBLEMS IN REACHING INTER NATIONAL AGREEMENTS ON IPR W h y are there so fe w (and such weak) interna tional agreements on IPR? One prim ary difficul ty in protecting and enforcing IPR is that the in centives to do so d iffer across countries, par ticularly betw een countries that are technology exporters and those that are technology im porters.24 Generally, less innovation occurs in developing countries; instead, in the absence o f licensing or foreign direct investment, firm s in these countries tend to produce goods whose production technology has becom e standardized or whose patent protection has expired. Firms that successfully pirate technology in many developing countries are often able to produce essentially the same product at sub stantially low er production costs. Because there is less innovation in developing countries, the cost o f not protecting IPR (reduced future inno vation) is often less, at least in a static sense, than the gain associated w ith selling these products. An argument often made by developing coun tries is that there is "excessive” protection o f IPR in industrial countries. The optimal amount o f protection o f intellectual property is difficult to determine w ithin a given country. This issue becomes m ore complicated in an international context, because w hat is optimal from a domes tic perspective may not be optimal from an in ternational standpoint. For example, even from a long-run perspective, the optimal amount o f IPR protection can d iffer across innovating and non-innovating countries.25 desired amount of protection between innovating and non innovating countries depends on the specification of social welfare used and that strong IPR may not improve global efficiency. Diwan and Rodrik (1989), in a different theoretical framework, also find that the optimal amount of protection between the innovating and non-innovating countries coincide only if welfare in the two regions is weighted equally. 45 Many poorer developing countries find it dif ficult to pay the higher price innovated goods w ould cost if they are given patent protection in their country. Developing countries th erefore have little incentive, at least in the short run, to protect commodities they need (or want) but could not afford to buy if protection is aw ard ed. An example o f this is seen in the pharma ceutical sector, which is awarded patent protec tion in essentially all industrial countries, but not in many poorer developing countries, which might otherwise have difficulty purchasing medical supplies. Another argument fo r not protecting IPR in technology-importing countries is that reverse engineering, the process by which firms take products apart to learn how to produce them, enables firms to learn how to develop new pro ducts themselves and th erefore aids in a coun try ’s development. As the technological kn ow how im proves in a country, these firm s begin to innovate themselves. W hen this stage is reach ed, protection o f IPR generally begins to in crease.26 Given the current level o f protection o f IPR, creating increased standards o f protection, at least in the short run and fo r the least-developed countries probably in the long run, w ill redistribute income from technology-importing, developing countries (whose residents w ill now have to pay m ore fo r these types o f products) to countries w hich innovate or already protect IPR. As a result, success in negotiating increased international protection fo r IPR w ill likely re quire some concessions to the developing coun tries in other areas o f trade.27 These issues are currently being discussed by W IPO and GATT. CONCLUSION The incentive to protect intellectual property rights, as w ell as the actual amount o f protec tion awarded, differs w idely across industrial and developing countries. Nevertheless, recent 26This has recently occurred in Korea, which, according to a 1987 study by the International Trade Commission, was the third-largest source of U.S. losses resulting from viola tions of IPR. Recently, however, Korea has taken steps to improve its enforcement of IPR, and the number of patents issued in the United States to Korean nationals has risen substantially, although it is still small in absolute value (U.S. Department of Commerce, Patent and Trademark Of fice, 1989). Of course, U.S. pressure has provided addi tional impetus for Korea to increase IPR protection. negotiations under W IPO and G A TT (which, as o f this writing, is yet to be concluded) regarding the trade-related aspects o f intellectual property rights indicate increased support fo r interna tional agreements on intellectual property rights and a realization that such agreements benefit both industrial and developing countries. In ad dition, as the amount o f inventive activity and the number o f countries engaged in innovation increases, the trend tow ard m ore cooperation and protection o f intellectual property is likely to continue. REFERENCES Benko, Robert P. “ Intellectual Property Rights and the Uruguay Round,” World Economy (June 1988), pp. 217-31. _______ Protecting Intellectual Property Rights: Issues and Controversies (American Enterprise Institute for Public Policy Research, 1987). “ Bugs that Divide.” Economist, July 28, 1990, pp. 57-58. Butler, Alison. “ Endogenous Innovation in a North-North Model of the Product Cycle,” Federal Reserve Bank of St. Louis Working Paper 90-007 (1990). Chin, Judith C., and Gene Grossman. “ Intellectual Property Rights and North-South Trade,” Princeton University Discus sion Papers in Economics, No. 143 (November 1988). Congress and the Nation. Vol. 6 (Congressional Quarterly Inc., 1985), p. 548. Diwan, Ishac, and Dani Rodrik. “ Patents, Appropriate Technology, and North-South Trade,” National Bureau of Economic Research Working Paper No. 2974 (May 1989). Feinberg, Robert M. “ Intellectual Property, Injury, and Inter national Trade,” Journal of World Trade (April 1988), pp. 45-56. Feinberg, Robert M., and Donald J. Rousslang. “ The Economic Effects of Intellectual Property Right Infringe ments,” Journal of Business (January 1990), pp. 79-90. Finger, Michael J., and Andrzej Olechowski, eds. The Uruguay Round: A Handbook on the Multilateral Trade Negotiations (The World Bank, 1987). General Agreement on Tariffs and Trade. GATT: What It Is, What It Does (1990). Grossman, Gene M., and Elhanan Helpman. “ Product Development and International Trade,” Journal of Political Economy (December 1989), pp. 1261-83. Hood, Neil, and Stephen Young. The Economics of Multina tional Enterprise (London: Longman Group United, 1979). 27For a review of some of the problems in negotiating IPR in the Uruguay Round, see Maskus (1990), Benko (1988) and Finger and Olechowski (1987). NOVEMBER/DECEMBER 1990 46 Kamien, Morton I., and Nancy L. Schwartz. Market Structure and Innovation (London: Cambridge University Press, 1982). Mansfield, Edwin. “ Patents and Innovation: An Empirical Study,” Management Science (February 1986), pp. 173-81. Maskus, Keith. “ Normative Concerns in the International Pro tection of Intellectual Property Rights,” paper prepared for meetings of the International Seminar in International Trade at the National Bureau of Economic Research (Auqust 2-3, 1990). _______. “ Intellectual Property,” in Jeffrey J. Schott, ed., Completing the Uruguay Round: A Results-Oriented Ap proach to the GATT Trade Negotiations (Institute for Interna tional Economics, 1990), pp. 164-79. Meessen, Karl M. “ Intellectual Property Rights in Interna tional Trade,” Journal of World Trade Law (February 1987), pp. 67-74. Nordhaus, William D. Invention, Growth, and Welfare: A Theoretical Treatment of Technological Change (MIT Press, 1969). “ Patents for Pharmaceuticals.” Financial Times, May 15, 1990 FEDERAL RESERVE BANK OF ST. LOUIS Plock, Ernest. “ International Piracy of Motion Pictures,” in Ann Main, “ Pursuing U.S. Goals Bilaterally: Intellectual Property and ‘Special 301’,” Business America (September 25, 1989), p. 7. U.S. Department of Commerce, Patent & Trademark Office. Commissioner of Patents and Trademarks Annual Report (January 1989). United States International Trade Commission. “ Foreign Pro tection of Intellectual Property Rights and the Effect on U.S. Industry and Trade,” Report to United States Trade Representative, Investigation No. 332-245, Publication 2065 (February 1988). World Intellectual Property Organization. General Information pamphlet (Geneva, January 1990a). _______ . WIPO Newsletter (February 1990b). _______. Existence, Scope and Form of Generally Interna tionally Accepted and Applied Standards/Norms for the Pro tection of Intellectual Property (Geneva, September 1988). Yuetter, Clayton. News Release. “ Statement by Ambassador Clayton Yuetter Concerning Intellectual Property Rights Protection,” Office of United States Trade Representative (April 7, 1986). 47 Michelle R. Garfinhel Michelle R. Garfinkel is a senior economist at the Federal Reserve Bank of St. Louis. Scott Leitz provided research assistance. The Economic Consequences of Reducing Military Spending B EFORE THE CRISIS in the Middle East this summer, the easing o f international tensions had reduced, fo r many, the urgency fo r the United States to continue building or even main tain its military strength. As support fo r allo cating the nation’s resources to defense w eak ened, people began to argue about the potential fo r a significant "peace dividend” available to the U.S. econom y.1 H ow should the savings from reduced defense spending be put to use to enhance our nation’s w elfare? Many analysts, concerned primarily about the effects o f large public deficits, have argued that the savings should be applied to reduce the governm ent’s need to b o rrow .2 Others have voiced concern that a reduction in defense expenditures w ill generate unem ploy ment, at least tem porarily, w hile resources are reallocated to productive activities in the civilian sector.3 Consequently, they have argued the in itial savings should be used to ease this adjust ment—perhaps by increasing expenditures on 'For example, see Pennar and Mandel (1989). 2See, for example, Schultze (1990). 3For example, see “ Peace Dividend or Recession?” (1990). Also see Pennar and Mandel (1989), who report the results of a study of the short- and long-term effects of reducing the defense budget by 5 percent a year in real terms from 1991 to 1994 and keeping it constant thereafter. Although this study predicts enhanced economic growth in the long run, it also predicts some short-term losses. Also see Ellis training programs.4 Many other policy recom mendations have been made. Although the current situation in the Middle East raises doubt that there w ill be any signifi cant dividend in the near term, it does not de tract from the relevance o f such recom menda tions. Instead, it provides us with m ore time to evaluate the various options associated with future defense cuts. In review in g the economic implications o f reduced military spending, this article examines some issues that have been overlooked b y those in search o f ways to use the “peace dividend.” The article begins w ith a b rief analysis o f re cent trends and the prospects fo r future cuts in military spending to see how large a dividend might be. Some simple economic principles are then em ployed to assess how the peace dividend, regardless o f its actual magnitude, might be used to achieve diverse economic goals. and Schine (1990), who report the results of a study in dicating that as many as 1 million defense-related jobs (or 20 percent of all jobs in defense-related activities) could be lost by 1995. Although other analysts have argued that the employment losses could be insignificant (see, for exam ple, Uchitelle (1990)), some defense contractors have already begun to cut production and employment. 4See, for example, the bill proposed recently by Senator Pell (S.2097). NOVEMBER/DECEMBER 1990 48 decades. Its share in the 1980s was only onethird o f that in the 1940s. Table 1 Trends in U.S. Defense Spending in the Past 50 Years Total defense spending1 Decade 1940s 1950s 1960s 1970s 1980s Nominal Real $ 35.8 40.2 59.2 90.3 238.3 $285.5 187.0 216.4 200.0 274.4 Share of GNP 18.3% 10.1 8.4 5.8 6.1 'Billions of dollars averaged over the decade. The real figures are in 1989 dollars. H O W BIG MIGHT THE DIVIDEND BE? T o place the discussion o f the economic im plications o f reduced defense needs into perspec tive, it is helpful to examine recent patterns o f m ilitary spending. Table 1 shows three v e ry dif feren t perspectives o f U.S. m ilitary spending over the past 50 years. W e can see that nominal m ilitary spending in the United States has grow n considerably since the 1940s, rising from an average o f $35.8 billion per year during the decade that included W o rld W a r II to an average o f nearly $240 billion per year during the 1980s. A fte r adjusting fo r inflation, how ever, w e see a somewhat different picture. Real military spending declined sharply im m ediately after the W W II decade; and, its pat tern since the 1950s has been m ore erratic, w ith its net rise by the 1980s being considerably less dramatic than suggested by the nominal figures. Finally, as the table shows, military spending as a fraction o f gross national product (GNP) has fallen m arkedly over the past five 5During the Reagan Administration’s military build up, defense spending grew at a 5.0 percent annual rate after adjusting for inflation; in comparison, the annual growth rate in real GNP over this same period was only 3 per cent. Earlier this century, before WWI, military spending’s share of GNP was less than 1.5 percent. (Because data on military spending before the 1940s are not entirely consis tent with the data presented in table 1, they are not shown here.) 6NATO’s proposal then called for a reduction of its own and the Warsaw Pact’s ground capabilities in Europe and for a reduction in their tactical aircraft capabilities that would leave an advantage for NATO. (Ground capability is mea sured by army units intended to fight on ground. Tactical aircraft capabilities are measured by fighters and bombers http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis W hile m ilitary spending might now represent a fairly small proportion o f GNP, it is still an im portant component o f economic activity. Defense spending was $301.1 billion in 1989—m ore than $1200 per person in the United States.5 This number suggests that there could be a consider able dividend from a large-scale disarmament. Even without the Middle East crisis, how ever, the military spending cuts likely to have occurred in the near future w ould have been quite small relative to the w hole economy. In a recent study, fo r example, the U.S. Congressional Budget O f fice (1990) estimated the effects o f a proposal by the North Atlantic Treaty Organization (NATO) fo r limiting conventional forces in Europe.6 It found that the treaty w ould generate an annual savings o f about $3 billion in 1990 prices. Although $3 billion seems large, it constitutes less than 1 percent o f the total Defense Departm ent’s budget authority fo r fiscal year 1990 and less than .06 percent o f 1989 GNP. This amounts to less than $13 per year per U.S. citizen. This estimated savings from the prospective reduction in m ilitary spending pales in compari son to earlier U.S. disarmament efforts fo llow ing w artim e periods.7 A fter W orld W a r II, fo r example, defense spending fell by about $57.3 billion from 1945 to 1946, almost 27 percent of GNP in 1945. From 1953 to 1954, after the K ore an War, defense spending fell b y $7.4 billion, almost 2 percent o f GNP in 1953. Forecasts o f the actual size o f future defense cuts, o f course, are subject to much uncertain ty. The budget proposed b y President Bush in January 1990 fo r fiscal year 1991 called fo r reducing the defense budget by 2 percent after intended to fight in the air and on the ground using con ventional weapons.) It called for a 27 percent reduction in a selection of NATO weapons. Assuming an equally pro portionate reduction by all NATO members, this require ment implies that the U.S. would have had to remove and destroy 600 tanks, 122 armored personnel carriers, 112 pieces of artillery, 189 helicopters and 105 aircrafts from Europe. Furthermore, about 30,000 U.S. troops would have had to be withdrawn from Europe and demobilized. 7As noted below, however, the economic implications of a given reduction in military spending depend quantitatively on whether the reduction follows a war or occurs during a relatively peaceful period. 49 adjusting fo r inflation over the fiscal years 1991 to 1995.8 Others argued fo r defense spending cuts as much as 5 percent in real terms per year from fiscal years 1992 to 1994, achieving an annual savings o f about $60 billion (in 1989 prices) starting in 1994.9 If cuts o f this magni tude w ere implemented, the implied savings w ould constitute about 1.2 percent o f GNP fo r the year 1989, nearly $250 per year on a per capita basis. Figure 1 PPC: Defense and Nondefense Goods And the savings could be even larger. Indeed, on the day o f the Iraqi invasion o f Kuwait, President Bush announced that, although the in vasion indicates a need to maintain a strong military force, U.S. armed forces could be reduced b y 25 percent over five years given the recent changes in Soviet-U.S. relations.10 EFFECTS OF THE DIVIDEND: SOME BASIC ECONOMIC PRINCIPLES Generally speaking, the trade-off betw een competing uses o f resources implies that a re duction in real m ilitary spending w ould provide a "dividend” in the fo rm o f increased real private and public consumption and investment oppor tunities—that is, increased resources available fo r the production o f consumption and invest ment goods. T o quantify these increased oppor tunities over time, the annual peace dividend is defined here simply as the annual reduction in real military spending.11 Figure 1 illustrates this trade-off given the na tion's resource and technology constraints by means o f a hypothetical production possibility curve (PPC) fo r defense goods (national security) and nondefense goods (public and private in vestment and consumption). This curve depicts the maximum quantities o f defense goods (M) and nondefense goods (N) that can be produced simultaneously fo r given amounts o f capital and labor inputs. s“ Peace Dividend or Peace Recession?” (1990). 9“ The Peace Economy” (1989). 10Dowd (1990). Specifically, the Pentagon plan then called for cutting the armed forces by 500,000 troops from the current level of 2.1 million. But, without a clear resolution of the ongoing conflict in the Middle East, any reduction in military spending might seem optimistic. Assuming that resources are fully utilized, the econom y is always operating on the fron tier regardless o f the level o f military spending. If no resources w ere allocated to the production o f defense goods, fo r example, the total output o f nondefense goods w ould be shown as N0. Producing defense goods thus requires sacrific ing the production o f some goods fo r invest ment and consumption. The opportunity cost o f providing a specific level o f national security, fo r example, MA, is the value o f the lost produc tion o f nondefense goods, NA- N o. Conversely, if the econom y w e re originally operating at point A, cutting military spending out entirely would imply increased production o f nondefense goods, N a - N c. Thus, w ith a reduction in military spend ing (say M B- M A), the annual dividend w ould im ply increased opportunities fo r the production o f nondefense goods (NB- N A). The shaded insert on pages 50 and 51 contains a discussion o f the w elfa re implications o f increased opportunities fo r investment and consumption afforded by a reduction in military spending. 11This definition envisions the dividend as a flow—i.e., as the term is normally understood. Thus, a permanent cut of X dollars in real terms per year implies an X dollar divi dend each year indefinitely into the future. Using a present discounted value concept, these flows over time can be translated into a stock concept: X/r, where r is the cons tant real interest rate. NOVEMBER/DECEMBER 1990 50 W ill a Dividend Necessarily Enhance “Social Welfare?” Some interesting social w elfare implications o f low er military spending can be illustrated within the simple PPC fram ew ork. Figure 2 depicts the same PPC shown in figure 1; add ed to the figure are tw o indifference curves that show specific combinations o f defense and nondefense goods w hich yield constant levels o f national “w elfa re” or "utility." The curve labeled u indicates a higher level o f utility than the one labeled u'. T h e shape o f the indifference curves reflects the notion o f diminishing social marginal utility. For exam ple, the nation is “less w illing” (must be given considerably m ore nondefense goods) to decrease consumption o f defense goods and consume m ore nondefense good at point Y compared w ith point Z. At the point o f tangency (labeled X) be tw een the indifference curve associated with utility u and the PPC, the nation’s utility is maximized, given the available resource and technology constraints. At this optimum level o f consumption, the marginal utility trade-off betw een defense and nondefense goods (the slope o f the indifference curve, u) equals the marginal rate o f transform ation betw een them (the slope o f the PPC). Th e optimal quantities o f defense and nondefense goods are M x and Nx, respectively. If preferences remain unchanged, a m ove to any other p ro duction combination on the PPC (or inside) w ould result in a low er level o f national w elfare. Thus, although a m ove from point X to point Y, fo r example, produces a positive dividend in the form o f increased resources available to produce nondefense goods, it w ould actually reduce the nation’s w elfare. Th ere are tw o scenarios, how ever, under w hich a decline in military spending could enhance w elfa re by creating increased con sumption and investment opportunities. First, the original quantities o f M and N could have been suboptimal. If, fo r example, the econ om y w ere operating at a point to the left o f X on the PPC (for example, point Z), military spending w ould be considered too high from Figure 2 Figure 3 Optimal Levels of Defense and Nondefense Goods A Preference-Induced Change in the Optimal Levels of Defense and Nondefense Goods FEDERAL RESERVE BANK OF ST. LOUIS 51 a social w elfa re point o f view. Alternatively, the original point could have been inside the PPC above point X, such as Z'. In this case, military spending is also too high and, m ore important, the econom y is operating below its potential output levels. Second, a cut in military spending could generate w elfa re gains if it w e re associated w ith an unanticipated decline in the severity o f external threats. This seems to be the ex ample most relevant to the current situation involving U.S.-Soviet relations. The reduced value o f maintaining existing U.S. military strength in the face o f detente can be cap tured best by presuming there has been a The concave shape o f the PPC reflects dimin ishing marginal returns in productive transfor mation. That is, the amount o f nondefense goods that must be sacrificed to produce one m ore defense good increases as M increases. For example, in the figure, the m ove from C to A and the m ove from A to B involve identical reductions in military spending. Starting at the point w ith a higher level o f m ilitary spending (Mc), how ever, that reduction in military spend ing implies greater additional production o f non defense goods than w hen starting at point B. (That is, N a - N c > N b - N a.) Hence, assuming resources are fully employed, a given reduction in the production o f defense goods w hen the current level is low —fo r example, during peacetime—w ould imply smaller additions to consumption and investment opportunities than when the current level is high—fo r example, during wartime. THE CROWDING-OUT EFFECT OF MILITARY SPENDING AND THE LONGER-RUN EFFECT OF THE PEACE DIVIDEND These allocative effects o f the peace dividend can also have implications fo r the amount o f resources available fo r production in general. ,2Of course, one might argue that the level of military spen ding could positively influence the position of the PPC. Because military spending enhances a nation’s ability to protect its resources, it might serve to increase the na tion’s future resource base by encouraging more investment. general clockwise m ovem ent in the indif ference curves. This m ovem ent is illustrated in figure 3, w h ere the new indifference curves are labeled u. I f national security can be maintained w ith less m ilitary spending, a smaller quantity o f M with the same quantity o f N (Nx) produces the same utility level uo= uo. Further, the marginal value o f increas ing M relative to the cost o f foregone con sumption o f N falls. The new w elfa re max imizing combination o f M and N along the PPC is denoted by Y, w h ere N increases to NY and M decreases to M Y. The net "w elfare e f fect” o f the dividend, M Y- M X, w ould be reflected in the increase from u to u;. Specifically, any additional investment adds to the future resource base, thereby enhancing future output grow th and investment and con sumption opportunities. Hence, even if the de cline in military spending w ere tem porary, fo r example lasting only one year, its benefits in terms o f increased productive capacity could be realized over many years. This longer-term e f fect can be m odeled in the fram ew ork presented above by an outward shift o f the PPC curve over time.12 Some analysts believe that, in recent decades, military spending has been excessive, reducing the residual supply o f productive resources (that is, capital and labor) available fo r private and nondefense public investment and thereby w eak ening the econom y.13 According to this "deple tion” theory, the effects o f higher levels o f military spending are reflected in low er rates o f investment and, consequently, low er rates o f economic growth. Thus, the principal result o f reduced military spending w ould be greater in vestment and econom ic growth. Evidence on the "crow ding-out” effect of military spending is based prim arily on empirical analyses relating changes in military spending’s share o f GNP to the GNP shares o f other broad categories o f expenditures; typically, the studies focus on the effect on private investment’s 13See, for example, Dumas (1987), Melman (1988) and Du Boff (1989). In contrast, Weidenbaum (1990) argues that, as defense spending has fallen relative to GNP, the effects of such spending on the U.S. economy have become less significant. NOVEMBER/DECEMBER 1990 52 Table 2 Broad Categories of Real Expenditures as a Share of Real GNP1 Decade 1940s 1950s 1960s 1970s 1980s C 54.7% 58.4 59.5 62.2 64.5 1 12.1% 16.5 16.7 17.2 16.9 X Gnm Gm 10.1% 11.8 14.3 14.9 13.6 23.0% 13.2 10.3 6.2 6.2 0.5% 0.1 -0.4 -0.7 -1 .2 ’ Each category is converted into real terms using its im plicit price deflator. Separate price deflators were used for federal government spending and state and local government spending, but both defense and non-defense federal expen ditures were deflated by the same number. share o f GNP.14 Table 2 shows the trends in personal consumption expenditures (C), private domestic investment (I), nondefense public ex penditures (Gnm) on goods and services, defense expenditures (G J and exports net o f imports (X) as shares o f GNP over the past 50 years. To focus on the real effects o f military spending, each broad category o f spending is converted into real terms b y dividing it b y its ow n price deflator and by real GNP.15 As the table suggests, real military spending has crow ded out all categories o f real expendi 14See Gold (1990) for an extensive survey of this evidence. Also see Adams and Gold (1987) and U.S. Congressional Budget Office (1983). It should be noted that identifying the degree to which military spending has resulted in lower expenditures on other goods and services and lower economic growth is difficult. The problem lies in determin ing how investment and other expenditures would have behaved if military spending had been different—in an ex treme case, if it had been zero. Because reduced-form parameters relating defense expenditures to other expen ditures would not be independent of the policy regime, estimates of these parameters might provide little informa tion on how a permanent change in military spending (i.e., a policy regime change) would influence other expen ditures. In addition, it is important to note that, if higher military expenditures result in higher levels of GNP, lower shares of investment, for example, need not imply a crowding-out effect of military spending. 15A comparison of tables 1 and 2 reveals that real military spending as a fraction of real GNP was higher than nominal military spending as a fraction of nominal GNP from the 1940s to the 1970s, but slightly lower during the 1980s. This divergence reflects the difference between the general price level of defense goods and that of all goods and services. The price level for defense goods, on average, was lower than the general price level between the 1940s and the 1970s, but higher during the 1980s. The focus on real military spending here is intended to FEDERAL RESERVE BANK OF ST. LOUIS tures, not just real private domestic investment. As real military spending’s share fell from the 1940s to the 1980s by 16.8 percentage points, real consumption’s share rose b y nearly 10 per centage points. Real private investment’s share and real nondefense public spending’s share also rose over this period, though less dramati cally.16 O f course, in a broad sense, the substitu tion observed betw een nondefense and defense public expenditures is consistent w ith the crowding-out notion; in this case, public invest ment on the nation’s infrastructure—that is, highways, airports, mass transit, w ater systems—was crow ded out.17 T o be sure, the size o f the peace dividend, as defined here, is independent o f its allocation among the production o f private and public consumption and investment goods. But its longer-term implications depend on that alloca tion. The best or "socially optimal” reallocation o f resources among investment and current con sumption depends on the nation’s willingness to forego current consumption in order to invest and thereby enhance future consumption possi bilities. The greater this willingness, the m ore likely the resources from a reduction in military spending w ill be devoted to additional invest ment rather than additional current consump tion. The low er the nation’s willingness to forego current consumption to enhance future consumption, the low er w ill be the proportion emphasize the importance of its real allocative effects. Failure to account for relative price movements masks these effects. 16Smith (1977) finds that the crowding-out effect of military’s share of income on investment's share of income is nearly one to one for 14 OECD nations during the 1960s. How ever, Boulding (1973), Edelstein (1989) and Aschauer (1989a) argue that this effect is not empirically relevant for the United States. Also see Browne (1989), who questions the validity of the argument that military research and de velopment has crowded out civilian spending on research and development by “ depleting” our nation’s scientists and engineers. As is well-known, military R&D has pro duced important innovations that have been applied suc cessfully to production activities in the civilian sector; one commonly cited example is the computer. In addition, Browne argues that this crowding-out effect on R&D pre sumes that the supply of scientists and engineers is fixed. A greater demand for highly skilled labor, however, has in fluenced its supply, although with the usual lag. 17As an annual average of GNP, net public infrastructure in vestment fell from approximately 2.3 percent in 1960-65 to about 0.8 percent in 1980-85 (Du Boff (1989), p. 7, table 2). Aschauer (1989b) argues that the recent reduction in public capital, including infrastructure, might be responsi ble for the recent decline in productivity. 53 o f the savings that is allocated to investment. Thus, in contrast to the suggestion o f the deple tion theory briefly described above, the divi dend from reduced military spending need not result in significantly greater rates o f invest ment and economic growth. The extent to w hich the dividend w ill affect economic grow th depends on h ow it is allocated among the p ro duction o f investment and consumption goods. ALTERNATIVE USES OF THE DIVIDEND In recent decades, decreases in nominal military spending typically have been associated with increases in other public expenditures in nominal terms. Indeed, after falling to 19.5 per cent in the 1950s from 25.1 percent during the 1940s, total public spending (federal plus state and local) on goods and services has remained roughly constant as a fraction o f nominal GNP, around 20 percent; only the composition o f those expenditures changed. Although there could be reasons w h y this pattern might persist in the upcoming decades, many analysts have questioned w hether a continuation o f this pat tern is either likely or even desirable. N ever theless, the basic question to be addressed should be couched in real terms; W hat should be done with the peace dividend? Increase Nondefense Public Expenditures Some analysts have argued that leaving nondefense public expenditures alone and using the reduction in military spending to either decrease the deficit or low er taxes is not the best use o f the peace dividend. Instead, many o f them believe that at least part o f the savings from arms reduction might best be used to in crease nonmilitary governm ent spending— specifically, to rebuild the nation’s infrastruc ture.18 In terms o f the economic fram ew ork above, this policy w ould shift out the PPC, in 18For example, see Du Boff (1989) and Melman (1988). 19For example, see Bolton (1966), especially pp. 37-41. 20See Barro (1981). In studying the output effects of govern ment expenditures in the United States, he distinguishes permanent from temporary components of military spen ding. He finds that the effect of increases in temporary military spending (essentially wartime expenditures) on output was nearly one-for-one; increases in permanent military spending also increased output, but by less than the change in military spending. 21See Barro (1981) for a theoretic discussion of the effects of government expenditures on output. In support of this creasing future production and consumption opportunities. Others have argued that, unless the fall in military spending is somehow offset, resource utilization and, hence, economic activity w ill fall as w ell.19 In this view , w hich lies w ithin the standard "Keynesian" paradigm, the governm ent should use part (or all) o f the savings to finance additional public expenditures, including noninfrastructure expenditures, such as w elfa re p ro grams, to offset the negative effect o f reduced military spending on aggregate demand. This argument assumes that military spending in particular or public expenditures in general enhance social w elfa re not only by providing additional public goods, but by increasing em ployment and thereby stimulating the econo m y—that is, by inducing the use o f idle resources. It implies that, w ithout the increases in military spending or, m ore generally, public expenditures during the post-W W II period, the econom y would have operated below its potential output capability (that is, inside its PPC). Although there is evidence that a permanent decrease in military spending can produce a permanent decline in aggregate output,20 the decline in output could be generated, in part, by a voluntary reduction in the supply o f labor. In other words, this evidence does not necessari ly imply that a permanent decline in military demand, without an increase in other public spending, w ould cause these productive re sources to becom e involuntarily idle on a per manent basis. Reduce Taxes Some analysts w ould like us to consider an alternative policy that leaves other governm ent expenditures unchanged and uses the dividend to reduce taxes. By increasing individuals’ after tax income, this policy might induce individuals to decrease their supply o f labor, which w ould in turn decrease output without leaving labor resources involuntarily idle.21 line of reasoning, Dunne and Smith (1990) find that for the United States, military spending does not “ cause” unemployment. Riddell (1988) argues, in a more Marxian spirit, that the government's apparent bias for military over non-military expenditures is driven by its objective to main tain international order so as to maximize the profitability of U.S. capital. This possible endogeneity of military spen ding calls many empirical analyses that treat military spen ding as exogenous into question. Also see Garfinkel (1990a), who uses a game-theoretic model to show how military spending can be driven by aggregate economic activity through the government’s motive to prevent other nations from extracting its citizens’ resources. NOVEMBER/DECEMBER 1990 54 It should be noted that a permanent decline in measured output, triggered by the impact o f reduced military spending (and reduced taxes) on leisure, does not necessarily reflect a deterio ration in social w elfare. Because leisure has value, w elfa re could increase even if consump tion did not. Further, the theory described above suggests that, although individuals would w ork less, they might actually consume m ore nondefense private goods because their disposable income has increased (their tax liabilities have declined). On net, their w elfare w ould have increased as long as this new out come w ere chosen voluntarily.22 Reduce the Public Deficit Some analysts, w ho view large public deficits as harmful to the economy, have argued that the governm ent should use the peace savings to reduce the public deficit.23 In particular, the large deficits (public dissavings) o f the past decade are thought to have caused a decline in total national savings—that is, the sum o f private and public savings. Since a decline in total savings decreases the residual supply o f credit available to private borrow ers, large public deficits are considered by many to have pushed up expected real interest rates (interest rates adjusted fo r expected inflation).24 Thus, using the dividend to reduce the public deficit w ould decrease expected real interest rates and thereby stimulate both investment activity and the production o f goods, such as exports and n ew homes, w hose sales are sensitive to movements in interest rates. Although the U.S. savings rate appears to have declined in recent years, how much o f this decline can be blam ed on large public deficits is 22As discussed below, however, labor and capital resources could be left involuntarily idle temporarily as the economy adjusts to the reduced military demand. 23See, for example, Schultze (1990). Also, see Chrystal and Thornton (1988) for a related discussion of the effects of deficit spending. 24lndeed, this effect on interest rates is thought to be the mechanism through which military spending has crowdedout investment. By enhancing the productivity of private capital, however, military spending could have had a “ crowding-in” effect that would have offset its crowdingout effect. But Aschauer (1989a,b) presents evidence that does not support the notion that additions to the stock of military capital add to the productivity of private capital. Moreover, the references cited in footnotes 14 and 16 pro vide evidence that military spending does not crowd out private investment. 25Schultze (1990) estimates that national savings as a percentage of national income (i.e., net national product) has fallen from an average of 8 percent during the three FEDERAL RESERVE BANK OF ST. LOUIS unclear.25 The argument that public deficits in fluence the national savings rate is based on a number o f potentially questionable assumptions. One is that individuals do not view tax cuts that increase public borrow in g (holding the level o f governm ent spending constant) as increasing their future tax liabilities. Instead, individuals feel w ealthier and increase their consumption in response to such tax cuts. Although they also might respond by increasing their savings, the increase in private savings is assumed to be in sufficient to keep total savings from falling. In this view , fo r a given level o f governm ent ex penditures, public deficits decrease total savings and increase aggregate demand.26 D oes the Timing o f Taxes Matter? Other analysts argue that individuals believe reductions in current taxes associated w ith addi tions to public debt must be financed eventually with additional future taxes. In light o f the in crease in their future tax liabilities, individuals increase their savings. Conversely, they respond to a decrease in the public deficit, fo r a given level o f governm ent expenditures, by decreasing their savings. In either case, consumption is unaffected. In this view , often referred to as the “Ricar dian" view , individuals behave as if a decrease in the public deficit results in an equal decrease in the present discounted value o f their future tax liabilities. This argument builds on the assumption that public debt w ill be retired eventually out o f future taxes. I f this view is valid, private and public savings fo r a given level o f governm ent expenditures should be perfectly negatively correlated, w hile total sav ings should be unrelated to public savings.27 decades before 1980 to 3.3 percent during the first three quarters of 1989. Some analysts, however, question the notion that savings is too low in the United States; their skepticism is based on problems with the conventional measurements of savings. Cullison (1990) provides a useful survey of this literature. 26See Thornton (1990) for a theoretical discussion of the link between total national savings and public deficits. 27Against this Ricardian view, one might argue that deficits could be financed either through increased seigniorage or income taxes in future generations. In either case, the “ burden” of the current deficit could be shifted and budget deficits, holding government expenditures fixed, could affect economic activity. See Barro (1989) for a brief discussion of the empirical evidence on the effects of budget deficits. While recognizing the problems associated with testing the Ricardian proposition, Barro argues that the existing evidence lends more support to the Ricardian view than to the alternative view (p. 52). 55 To be sure, using part o f the annual dividend to reduce the public deficit could increase total investment and total consumption. According to the Ricardian view , how ever, the amounts o f these increases do not depend on w hether taxes are cut or the deficit is reduced. A cut in the deficit reduces future tax liabilities, but the tim ing o f the tax cuts does not matter. Because o f the distorting nature o f the income tax system, how ever, the equivalence betw een taxes and debt creation implied by the Ricardian view w ould be, at best, a rough approximation. Economic theory predicts that proportional in come taxation distorts individuals' decisions about consumption and labor supply. These distortions are costly and, other things being equal, the severity o f the distortion increases as the tax rate increases. Consequently, if the federal governm ent wants to minimize the costs associated with these distortions, given the path o f future governm ent expenditures, it should smooth income taxes over time.28 This m odified version o f the Ricardian view suggests that tax reductions, rather than deficit reductions, w ould be a preferable use fo r the peace divi dend.29 THE TRANSITIONAL COSTS OF ECONOMIC ADJUSTMENT In thinking about how the savings from re duced military spending could be used, it is im portant to consider how the econom y adjusts to unanticipated changes in resource uses. Reduced military spending w ill produce a negative shortrun effect on production, as labor and capital resources are shifted from military to civilian uses. During the transition period, some re sources w ill be unem ployed or underemployed. A Historical Perspective Previous disarmaments have been associated w ith sizable reductions in economic activity.30 28See Barro (1979). Also see Garfinkel (1990b) for an exten sion of this theory to include conscription as an additional tool for financing public expenditures to avoid the distor tions of income taxes, particularly during periods of severe military needs. 29ln this view, deficits are necessary to smooth out the distortionary effects associated with taxes. Hence, a tem porary increase in government spending should be financed with debt and a temporary decrease in government spend ing should result in a budget surplus; this tax-smoothing view of debt creation predicts that deficits are temporary phenomena. Although historical evidence supports this From the first quarter o f 1945 to the first quar ter o f 1946 (peak to trough), fo r example, nominal GNP fell at a seasonally adjusted an nualized rate o f $22.8 billion or 10.3 percent. These numbers understate the magnitude o f the decline in output as, during this period, there was a considerable acceleration in inflation.31 In real terms, GNP fell 18 percent over this period, and it was not until the third quarter o f 1952 that the level o f real GNP had fully recovered. Although this decline in real GNP is large, it is an overstatem ent o f the drop in national w el fare. W hile the level o f employment fell sub stantially, the unemployment rate rose very little. Instead, a substantial num ber o f workers, partic ularly wom en, voluntarily w ith d rew from the labor force. The transition to peacetime after W W II was facilitated, in part, by governm ent policies. Tax reductions and transfer payments (unemploy ment and veteran benefits) left disposable (net o f taxes) income nearly unaffected by the mas sive reduction in m ilitary spending. Thus, de mand fo r consumption goods rose to offset par tially the decline in military demand. The sharp decline in military spending that follow ed the end o f the Korean W a r was asso ciated w ith a mild recession. From the second quarter o f 1953 to the second quarter o f 1954 (peak to trough), real GNP fell 3.2 percent. De clines in defense spending and other federal governm ent expenditures, combined with inven tory decumulations, w e re the driving forces here. By the first quarter o f 1955, how ever, real GNP had climbed w ell above its previous peak. Factors Influencing The Adjust ment Costs The aggregate adjustment cost fo r any reduc tion in military spending can be measured by the real value o f resources (labor and capital) left idle involuntarily during the transition. For a given cut in defense spending, the magnitude positive theory of debt creation—see Barro (1979), for example—it is unclear whether the current deficit is only a temporary phenomenon. Indeed, the magnitude and per sistence of the peacetime deficits during the 1980s are un precedented in U.S. history. 30See Bolton (1966) and references cited therein for a more detailed examination of these periods of disarmament. Much of the discussion here draws from this work. Data are taken from Balke and Gordon (1986). 31This acceleration was driven, in part, by the removal of price controls in 1946. NOVEMBER/DECEMBER 1990 56 o f these costs depends on the speed w ith which labor and capital resources can be transform ed to meet new demands. The speed o f resource transformation, in turn, depends on the degree o f specialization o f resources used in the military sector. This spe cialization has tw o dimensions. First, certain in dustries, occupations and firms are highly de pendent on military demand. Second, military production is highly concentrated in several regions o f the United States. Such specialization w ill slow the adjustment process. A given reduction in military demand now might generate relatively greater adjustment costs than those associated w ith large-scale dis armaments follow ing wartim e periods. During w artim e periods, resources norm ally used to produce nondefense goods are m obilized quick ly and, presumably, on a tem porary basis; after the war, resources are rechanneled easily into their original civilian productive activities. In contrast, during peacetime defense firm s and their em ployees expect that demand fo r their product is essentially permanent. T o the extent that these firm s and employees have a compara tive advantage in the production o f defense goods, they are less likely either to diversify their operations into civilian markets or be able to do so in the event o f an unanticipated perma nent reduction in military spending.32 EVALUATING ALTERNATIVE PROGRAMS T o evaluate society’s options fo r using the sav ings from reduced military spending, w e must address tw o related economic issues. The first issue, already discussed, concerns w hat are the best uses o f the dividend or the new “ highestvalued” uses o f the resources previously used in the military (presumably their previous highestvalued uses). The second issue concerns how to rechannel resources efficiently from their mili tary uses to their new highest-valued uses. 32ln contrast, one might believe that defense contractors, having learned from past experience with sharp declines in military demand, would have diversified their operations to exploit commercial opportunities. While such diversifica tion would provide insurance against large losses to these firms and their employees in the event of an unexpected decline in military demand, past efforts in this direction have not been particularly successful. See Weidenbaum (1973) and Ellis and Schine (1990). “ Representative Boxer’s proposal (HR5327) is similar; however, it would penalize defense contractors who close FEDERAL RESERVE BANK OF ST. LOUIS Some people have advocated establishing public programs—fo r example, training programs—to lessen the costs o f adjustment borne solely by those closely linked to the m ilitary sector. A bill introduced recently b y Senator Pell (S.2097), fo r example, seeks to establish a program through which grants w ould be made to assist state and local governm ents in developing economic ad justment plans—fo r example, job retraining and finding alternative uses fo r defense facilities. These grants w ould be funded, in part, from the savings from reduced military spending.33 Anoth er bill introduced by Senator Coats (S.2682) is intended to aid defense contractors in diversify ing their operations into nondefense markets. Through tax incentives, this bill would encourage defense contractors and their employees to adopt em ployee stock ownership plans (ESOPs) to fi nance the corporate restructuring necessary to adjust to the reduction in military demand. These programs are aimed at distributing the adjustment costs and the benefits o f reduced military spending equitably; how ever, they are unlikely to effect an efficient reallocation o f resources.34 Consider, fo r example, a program that increases public nondefense expenditures on goods that are most easily produced b y the capital and labor resources originally em ployed fo r the production o f defense goods. W hile this program might w ell limit the adverse impact o f reduced military spending otherwise borne by those firm s and individuals highly dependent on military demand, it is clearly inefficient from the nation's point o f view. Unless increased spending on these other goods w ere deemed desirable on a permanent basis, this policy w ould not provide firm s and individuals with the incentives to channel their resources to new higher-valued uses. Instead, it w ould merely prolong the process o f adjustment and delay the realization o f the full benefits from a perm a nent reduction in military spending. Nevertheless, the redistributive effects o f reduced military spending should not be their firms unless a contract has been canceled. The states of Washington and California already have initiated adjustment plans. See Ellis and Schine (1990). “ Although the evidence on the effectiveness of manpower programs (for example, the Manpower Development and Training Act) is mixed, some studies find that manpower policies have been successful in raising the earnings of training program participants. See, for example, Ashenfelter (1978). 57 dismissed as unimportant or irrelevant in choos ing how the peace dividend ultimately w ill be used. The question o f w ho reaps the gains and w ho bears the costs o f an unanticipated reduc tion in military spending is an important aspect o f the problem and w ill play an important role in the solution. CONCLUSIONS This article has examined some possible e f fects o f a permanent reduction in military spen ding. In principle, the present discounted value o f the implied dividends from such a reduction, in terms o f increased consumption and invest ment opportunities, could be substantial. Through increased private and public invest ment, reduced military spending implies greater economic grow th and, hence, greater consump tion and investment opportunities in the future. The important economic questions are, How can these resources be reallocated efficiently to non military uses? and, H ow can w e identify what these uses should be? This article has introduc ed and discussed the economic issues that must be addressed in answering these questions; much further analysis and discussion w ill clear ly be needed before these questions can be answered adequately. O f course, given the recent course o f events in the Middle East, the reduction in military spending in the near future might not be large enough to generate any sizable dividend. Thus, debate over w hether the savings from reduced m ilitary spending should be used to reduce the public deficit, redistributed to taxpayers through tax cuts or be used to rebuild the in frastructure might seem premature. Because many communities, firms and individuals are af fected by even small reductions in military spending, how ever, the "m icro” costs o f these adjustments w ill not be ignored in the political decision-making process. But tem porary transitional costs do not justify abandoning the effo rt to reduce the amount o f resources allocated to military spending. M ore over, the Middle East situation is not a perma nent obstacle to realize a large dividend in the future. W hile cuts in defense spending are not expected to be particularly large now, a resolu tion o f the Middle East conflict w ill perm it a much larger dividend in the future. REFERENCES Adams, Gordon, and David Gold. Defense Spending and the Economy: Does the Defense Dollar Make a Difference? (Washington, D.C.: Defense Budget Project at the Center on Budget and Policy Priorities, July 1987). Aschauer, David. "Does Public Capital Crowd Out Private Capital?” Journal of Monetary Economics (September 1989a), pp. 171-88. _______ "Is Public Expenditure Productive?” Journal of Monetary Economics (March 1989b), pp. 177-200. Ashenfelter, Orley. “ Estimating the Effect of Training Pro grams on Earnings,'"Review of Economics and Statistics (February 1978), pp. 47-57. Balke, Nathan S. and Robert J. Gordon. "Historical Data,” in Robert J. Gordon, ed., The American Business Cycle: Con tinuity and Change (Chicago: The University of Chicago Press, 1986), pp. 781-850. Barro, Robert J. “The Ricardian Approach to Budget Deficits,” Journal of Economic Perspectives (Spring 1989), pp. 37-54. _______ “Output Effects of Government Purchases,” Jour nal of Political Economy (December 1981), pp. 1086-121. _______ “On the Determination of the Public Debt,” Journal of Political Economy (October 1979), pp. 940-71. Bolton, Roger E. “ Defense Spending: Burden or Prop?” in R.E. Bolton, ed., Defense and Disarmament (Prentice-Hall, Inc., 1966), pp. 1-53. Boulding, Kenneth E. “The Impact of the Defense Industry on the Structure of the American Economy,” in Bernard Udis, ed., The Economic Consequences of Reduced Military Spending (Toronto: D.C. Heath and Company, 1973), pp. 225-52. Browne, Lynn E. “ Defense Spending and High Technology Development: National and State Issues,” New England Economic Review, Federal Reserve Bank of Boston (September/October 1988), pp. 3-22. Chrystal, K. Alec, and Daniel L. Thornton. “The Macroeconomic Effects of Deficit Spending: A Review,” this Review (November/December 1988), pp. 48-60. Cullison, William E. “ Is Saving Too Low in the United States?” Federal Reserve Bank of Richmond Economic Review (May/June 1990), pp. 20-35. Dowd, Maureen. “ Bush Says Military Can Be Cut 25% in 5 Years,” New York Times, August 3, 1990. Du Boff, Richard B. “What Military Spending Really Costs,” Challenge (September/October 1989), pp. 4-10. Dumas, Lloyd J. “ National Security and Economic Delu sion,” Challenge (March/April 1987), pp. 28-33. Dunne, Paul, and Ron Smith. “ Military Expenditure and Unemployment in the OECD,” Defence Economics: The Political Economy of Defence, Disarmament and Peace, Vol. 1, No. 1 (1990), pp. 57-74. Edelstein, Michael. “What Price Cold War: Military Expen ditures and Private Investment in the U.S., 1890-1980,” mimeo (January 1989). Ellis, James E., and Eric Schine. “Who Pays For Peace?” Business Week (July 2, 1990), pp. 64-70. Garfinkel, Michelle R. “Arming as a Strategic Investment in a Cooperative Equilibrium," American Economic Review (March 1990a), pp. 50-68. _______ “The Role of the Military Draft in Optimal Fiscal Policy,” Southern Economic Journal (January 1990b), pp. 718-31. NOVEMBER/DECEMBER 1990 58 Gold, David. The Impact of Defense Spending on Investment, Productivity and Economic Growth (Washington, D.C.: Defense Budget Project at the Center on Budget and Policy Priorities, February 1990). Melman, Seymour. “ Economic Consequences of the Arms Race: The Second-Rate Economy,” American Economic Review (May 1988), pp. 55-59. “ Peace Dividend or Peace Recession?” The Economist (April 7, 1990), pp. 31-32. Pennar, Karen, and Michael J. Mandel. “ The Peace Economy: How Defense Cuts Will Fuel America’s LongTerm Prosperity,” Business Week (December 11, 1989), pp. 50-55. Riddell, Tom. “ U.S. Military Power, the Terms of Trade, and the Profit Rate,” American Economic Review (May 1988), pp. 60-65. Schultze, Charles L. “ Using the Peace Dividend to Increase Saving,” Challenge (March/April 1990), pp. 11-17. Smith, Ron P. “ Military Expenditure and Capitalism,” Cam bridge Journal of Economics (March 1977), pp. 61-76. http://fraser.stlouisfed.org/ FEDERAL RESERVE BANK OF ST. LOUIS Federal Reserve Bank of St. Louis _______ “ Military Expenditure and Capitalism: A Reply,” Cambridge Journal of Economics (September 1978), pp. 299-304. Stevenson, Richard W. “ Suppliers Brace for Arms Cuts,” New York Times, June 18, 1990. Thornton, Daniel L. “ Do Government Deficits Matter?” this Review (September/October 1990), pp. 25-39. Uchitelle, Louis. “ Economy Expected to Absorb Effects of Military Cuts,” New York Times (April 15, 1990). U.S. Congressional Budget Office. Defense Spending and the Economy (Government Printing Office, February 1983). _______ . Budgetary and Military Effects of a Treaty Limiting Conventional Forces in Europe, Special Study (January 1990). Weidenbaum, Murray. “ Defense Outlays Have Declining Im pact on Economy,” Christian Science Monitor, February 7, 1990. _______ “ Industrial Adjustments to Military Expenditure Shifts and Cutbacks,” in Bernard Uris, ed., The Economic Consequences of Reduced Military Spending (Toronto: D.C. Heath and Company, 1973), pp. 253-87. Federal Reserve Bank of St. Louis Post Office Box 442 St. Louis, Missouri 63166 The Review is published six times per year by the Research and Public Information Department o f the Federal Reserve Rank o f St. Louis. Single-copy subscriptions are available to the public fr e e o f charge. Mail requests f o r subscriptions, back issues, or address changes to: Research and Public Information Department, Federal Reserve Rank o f St. Louis, P.O. Roy 442, St. Louis, Missouri 63166. The views expressed are those o f the individual authors and do not necessarily reflect official positions o f the Federal Reserve Rank o f St. Louis or the Federal Reserve System. Articles herein may be reprinted provided the source is credited. Please provide the Rank’s Research and Public Information Department with a copy o f reprinted material.