The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Announcement Federal Reserve Bank^^ of Chicago 2002 Conference on Bank Structure and Competition First Quarter 2002 perspectives 2 The electricity system at the crossroads—Policy choices and pitfalls 19 The aggregate effects of advance notice requirements 32 When can we forecast inflation? 45 Origins of the use of Treasury debt in open market operations: Lessons for the present Conference on Bank Structure and Competition announcement Economic __ perspectives President Michael H. Moskow Senior Vice President and Director of Research William C. Hunter Research Department Financial Studies Douglas Evanoff, Vice President Macroeconomic Policy Charles Evans, Vice President Microeconomic Policy Daniel Sullivan, Vice President Regional Programs William A. Testa, Vice President Economics Editor David Marshall Editor Helen O’D. Koshy Associate Editor Kathryn Moran Production Julia Baker, Rita Molloy, Yvonne Peeples, Nancy Wellman Economic Perspectives is published by the Research Department of the Federal Reserve Bank of Chicago. The views expressed are the authors’ and do not necessarily reflect the views of the Federal Reserve Bank of Chicago or the Federal Reserve System. Single-copy subscriptions are available free of charge. Please send requests for single- and multiple-copy subscriptions, back issues, and address changes to the Public Information Center, Federal Reserve Bank of Chicago, P.O. Box 834, Chicago, Illinois 60690-0834, telephone 312-322-5111 or fax 312-322-5515. Economic Perspectives and other Bank publications are available on the World Wide Web at http:/www.chicagofed.org. Articles may be reprinted provided the source is credited and the Public Information Center is sent a copy of the published material. Citations should include the following information: author, year, title of article, Federal Reserve Bank of Chicago, Economic Perspectives, quarter, and page numbers. « es Chicago fed. org ISSN 0164-0682 Contents First Quarter 2002, Volume XXVI, Issue 1 The electricity system at the crossroads—Policy choices and pitfalls Richard Mattoon Can electricity markets be successfully opened to competition? Events ranging from California’s electricity crisis to the fall of electricity trading giant Enron have caused policymakers to reexamine the benefits of restructuring the industry. This article examines policy developments in the Midwest and highlights some lessons that might help guide future electricity policy. The aggregate effects of advance notice requirements Marcelo Veracierto This article analyzes the effects of advance notice requirements on aggregate output, wages, employment, and welfare levels. The author finds that, contrary to firing taxes, advance notice requirements do not lead to reductions in employment. However, they can reduce welfare levels considerably more than firing taxes. Conference on Bank Structure and Competition announcement When can we forecast inflation? Jonas D. M. Fisher, Chin Te Liu, and Ruilin Zhou This article reassesses recent work that has challenged the usefulness of inflation forecasts. The authors find that inflation forecasts were informative in 1977-84 and 1993-2000, but less informative in 1985-92. They also find that standard forecasting models, while generally poor at forecasting the magnitude of inflation, are good at forecasting the direction of change of inflation. Origins of the use of Treasury debt in open market operations: Lessons for the present David Marshall The Federal Reserve currently conducts open market operations primarily in Treasury securities. It has not always done so. In its earliest years, the Fed conducted open market operations primarily in private securities, such as bankers' acceptances. The Fed’s choice of instruments was based both on economic doctrine and to help foster a liquid secondary market in these securities. The move to reliance on Treasury securities resulted from changes in the financial markets and the prevailing economic doctrine. These historical antecedents may have relevance for current problems facing the Federal Reserve. The electricity system at the crossroads Policy choices and pitfalls Richard Mattoon Introduction and summary In the mid-1980s, electricity policy in the United States began a new chapter when wholesale electricity markets were opened to competition. While the immediate goal was to increase the diversity of supply for electricity generation, proponents of restructuring also cited other dimensions of success arising from the restructuring of other network industries (such as telecommunications, airlines, and natural gas) as justification for introducing competition to the electric utility industry. Wholesale competition for producing electricity would improve generation efficiency, diversify supply, promote innovation, and even lower prices. Success in opening the wholesale market, proponents argued, would eventually be extended to the retail market, and all consumers would have the opportunity to choose their supplier and pick an electricity service that best fit their individual needs. The initial enthusiasm for restructuring was particularly noticeable in states with high electricity prices. In theory, splitting the traditionally integrated functions of a utilitypower generation, transmission, and distributioninto separate functions would expose cross-subsidies and inefficiencies, and competition among power generators would lead to lower prices for all classes of customers. Restructuring was designed to introduce open market competition only in electricity generation. Transmission and distribution services would still be subject to varying levels of regulation. By 2000, almost half of the states were pursuing some form of restructuring. However, several recent events have cooled the enthusiasm for abandoning the traditional heavily regulated and integrated utility system. Foremost among these was the California electricity crisis. The state garnered daily headlines as a series of events, including a flawed restructuring plan, left California facing skyrocketing prices, potential blackouts, and bankrupt utilities. 2 Californias high-profile bad experience clearly demonstrated that the costs of a flawed electricity restructuring policy could be very high. In addition, states that had demonstrated early success in restructuring, such as Pennsylvania, Connecticut, and Massachusetts, were beginning to find that sustaining competition and promoting new market entrants was harder than they had anticipated. This apparent conflict between theory and outcome has left restructuring at a crossroads. States are examining what elements and structures need to be in place to realize the promise and benefits of opening electricity markets to competition. The questions policymakers need to answer include the following: ■ ■ ■ ■ Is the physical infrastructure (particularly, adequate supplies of generation and transmission) in place to support new market entrants and a competitive market? Are the incentives for investing in new electricity facilities adequate? What can be done to improve these incentives if they are lacking? Do new institutions need to be developed to facilitate this new structure for delivering electricity? Should these be federal, regional, state, or quasipublic institutions? What is the role for existing regulatory institutions? Should restructuring expose consumers to changes in electricity prices, even when those prices can be volatile? Richard Mattoon is a senior economist at the Federal Reserve Bank of Chicago. This research was conducted in conjunction with the Federal Reserve Bank of Chicagos Midwest Infrastructure and Regulation project. The author wishes to thank William Testa, Thomas Klier, and Jack Hervey for reviewing the manuscript. Able research assistance was provided by Margrethe Krontoft. 1Q/2002, Economic Perspectives ■ What is the relationship between meeting environmental goals and generating greater power supply? Can the two successfully coexist? In this article, I examine what restructuring means in the electricity field. I discuss the legacy of the existing electricity system, which favored local electricity provision by integrated and highly regulated monopoly utilities, and describe the issues involved in moving to a more market-based system. Then, I use the five states of the Seventh Federal Reserve District as a case study for examining how restructuring issues are being addressed at the state level. The states of the Seventh District provide a particularly useful example, given that restructuring programs in Illinois and Michigan are well underway with consumers to be provided with retail choice in 2002. In contrast, Indiana, Iowa, and Wisconsin have adopted a cautious approach to restructuring, as relatively low prices for electricity have led them to question the immediate benefits of abandoning their existing structure for delivering electricity. Based on this analysis, I identify some lessons that can be applied as electricity policy continues to evolve. Evidence suggests that defining the role of existing and new institutions in managing the transition to market competition is one of the keys to promoting electricity restructuring. This may include insulating these institutions from political interference. Similarly, we need to examine how markets are structured to provide access to competitive electricity supply sources, as well as recognizing how the unique attributes of electricity create challenges for trading power as a commodity. Finally, policymakers need to consider the role of the electricity consumer in restructuring. For restructuring to succeed, consumers need to be exposed and to respond to legitimate market-based changes in electricity prices. Price signals that reflect fundamental changes in the cost of generation need to be passed through to consumers. While consumers may be provided with tools to manage volatile electricity prices, creating barriers to prevent price changes from being reflected in utility bills will not provide incentives for consumers to conserve electricity or for firms to invest in expanded generation. Understanding the legacy of the U.S. electricity system For much of its history, the electric utility business has received little public attention. Electric policy assumed that utilities were natural monopoly providers of a regulated and essential public service. Consumers were told which company would be their electric provider and how much they would pay for the service based on the service territory they were located in. Federal Reserve Bank of Chicago Decisions regarding 1) how energy was generated, 2) if new plants were necessary, and 3) how much should be charged were largely discussed inside utility companies and in hearing rooms at state public utility commissions. There were good reasons for maintaining this structure. The electric utility business is a very capitalintensive industry. Investments in power plants, transmission, and distribution systems are expensive and long-lived, and it would be inefficient to build overlapping systems within the same service territories. The clear public policy response was to recognize the monopoly status of utility companies, provide the companies with defined geographic service territories, and then subject them to rigorous regulation so as to prevent the exercise of pricing power.1 The same rationale was applied to other network industries, such as telecommunications, where the policy goal of providing service to everyone (universal service) at a moderate price was viewed as a primary objective. For the most part, this led to a regulatory compact in which utilities received monopoly status in return for a pricing structure based on tariffs that were just and reasonable (for example, that reflected the utilities cost of production and delivery) and that provided for a fair rate of return on invested assets. This emphasis on local monopoly provision and local policymaking led to a highly fragmented electricity system in the U.S. Everything from the price charged for electricity to the fuel used for generation varied widely from region to region. Figure 1 demonstrates the extreme variability in the price (as measured in average revenue per kilowatt hour [kWh]). For example, while the price of electricity is a mere 4 cents in Idaho, where hydroelectric generation keeps costs low, it is nearly triple that amount in nuclear dependent New Hampshire at almost 12 cents. In both states, the average revenue received by the states utilities is justified based on a review by the state public utility commission of the cost borne by the utility to generate and deliver energy in its service territory. The choice of fuel is a very significant factor in price variability. Figure 2 provides a historical perspective on the costs of coal, natural gas, and petroleum at the national level. Coal has exhibited very steady and slightly declining costs, while petroleum and natural gas costs have demonstrated significantly more volatility. In particular, the rapid run-up in natural gas costs from 2000 through the first part of 2001 posed major challenges to natural-gas-fired generators. Electric prices also vary by class of customer served (see table 1). Industrial customers are often charged lower tariff rates because they are easier to 3 FIGURE 1 Average revenue from electricity sales to all retail customers by state, 1998 cents per kilowatt hour WA 4 ND 5.7 MT 4.8 OR 4.9 MN 5.7 MI 7.1 WY 4.3 UT 5.2 CO 6 IL 7.5 KS 6.3 CA 9 AZ 7.3 OK 5.4 NM 6.8 IN 5.3 MO 6.1 TX 6.1 LA 5.8 WV 5.1 TN 5.6 MS 6 AL 5.6 RI 9.6 PA 7.9 OH 6.4 KY 4.2 AR 5.8 MA 9.6 NY 10.7 IA 6 NE 5.3 NV 5.8 ME 9.8 WI 5.4 SD 6.3 ID 4 NH 11.9 VT 9.8 NJ DE 10.2 6.9 VA 5.9 NC 6.5 DC 7.4 CT 10.3 MD 7 SC 5.5 GA 6.4 0 to 5 5 to 6 FL 7 6 to 8 8 to 12 Source: U.S. Department of Energy, Energy Information Administration, 2000, “The restructuring of the electric power industry: A capsule of issues and events,” report, January. serve. As bulk users of electricity, they often draw a highly predictable and steady level of power and, as a result, their costs of service (for example, connections to the grid) are often lower than for residential customers. Providing residential service requires managing a more variable load FIGURE 2 and can only be accomplished through a U.S. electric utility average cost for fossil fuels, large distribution system, supported by 1990 through May 2001 higher maintenance and billing costs. The system of governance of utilities cost (cents/106 btu) 700 is also fragmented. For the most part large, vertically integrated, investor-owned 600 utilities are responsible for generating, transmitting, and distributing power to 500 customers. However, other forms of utility ownership are also popular, including 400 Gas municipal ownership, cooperative ownership, and even federal power utilities such 300 as the Tennessee Valley Authority and the 200 Bonneville Power Authority (see table 2). Petroleum These differences in governance have im100 Coal portant ramifications for regulatory outcomes. While large investor-owned utilities 0 (IOUs) are subject to review by state public 1990 ’91 ’92 ’93 ’94 ’95 ’96 ’97 ’98 ’99 ’00 ’01 utility commissions, many public power Note: btu is British thermal unit. authorities are exempt from these requireSource: U.S. Department of Energy, Energy Information Administration, 2001, “U.S. electric utility receipts of and average cost for fossil fuels, ments. This fragmented structure makes 1990 through May 2001,” table, available on the Internet at www.eia.doe.gov/ cneaf/electricity/epm/epmt26p1.html. electricity a policy area with many participants and little central planning or review 4 1Q/2002, Economic Perspectives in 1996 and were designed to pave the way for increased participation by non-utilities Electricity price by class of customer and promote wholesale competition by Value Highest Lowest eliminating local utility monopoly control over transmission. The combined effect of Average electricity these orders required public utilities that price (cents/kWh) 6.66 NH(11.75) ID(3.98) controlled transmission to develop open Industrial 4.43 NH(9.21) WA(2.70) access, non-discriminatory transmission Commercial 7.26 NH(11.39) ID(4.20) tariffs and to provide existing and potenResidential 8.16 NH(13.84) WA(5.10) tial users with equal access to transmission information. These orders also began the Note: Prices are based on the contiguous U.S. NH is New Hampshire; process of unbundling existing utility WA is Washington; and ID is Idaho. Source: U.S. Department of Energy, Energy Information Administration, functions by separating transmission of 1999, “Average revenue per kilowatt hour by sector, census division, electricity as a stand-alone service from and state (cents),” available on the Internet at www.eia.doe.gov/cneaf/ electricity/esr/t12.txt. generation and distribution.2 The opening of access to transmission lines was a significant step. States with high-priced authority, except within the balkanized areas served and electricity hoped that the development of an active and regulated by a public authority. open wholesale electric market would serve as a base for moving into retail deregulation. Increased wholeThe start of a new era— sale competition would provide local distribution comWholesale deregulation panies with more options over how to meet their load In 1978, the passage of the Public Utility Reguobligation, and eventually individual consumers would latory Policies Act (PURPA) opened the wholesale be able to choose their electricity generator. power market to certain non-utility generating comBy 1999, FERC pushed the issue of opening the panies. PURPA was passed to help reduce U.S. depentransmission grid one step further with the adoption dence on foreign oil and to expand the diversity of of Order 2000.3 This order encouraged states to form supply for U.S. electricity generation. By 1998, nonRegional Transmission Organizations (RTOs) to imutilities were responsible for 11 percent of the total prove the multi-state operations of the transmission generation in the nation and were contributing 406 grid. The RTO was to serve as a multi-state, indepenbillion kWh to the electric system. PURPA was foldent organization to manage the operation of the translowed by the passage of the Energy Policy Act of 1992 mission grid for particular regions. The order provides (EPACT). One aspect of EPACT was to further press specific (but voluntary) guidance concerning a miniwholesale deregulation by opening up transmission mum set of eight functions that an RTO must be able access to non-utilities. In return, regulated utilities to perform, but it leaves it up to the states and the utiliwere permitted to build new merchant plants outside ties to develop both the geographic footprint and the their service territories. governance structure of the RTO. The suggested eight Other landmarks in restructuring were regulatory minimum functions are: responsibility for tariff adOrders 888 and 889 issued by the Federal Energy Reguministration and design; congestion management; parlatory Commission (FERC). Both orders were issued allel path flow; ancillary services; total transmission TABLE 1 TABLE 2 Utility retail sales statistics, 1998 Number of utilities Number of retail customers Retail sales (mWh) Percentage of retail sales Investor-owned Public Federal Cooperative 205 1,951 7 852 91,889,360 18,002,349 33,544 14,115,259 $2,427,733,133 $485,692,301 $46,631,180 $279,761,845 74.9 15.0 1.4 8.6 Source: U.S. Department of Energy, Energy Information Administration 1999, State Electricity Profiles, available on the Internet at www.eia.doe.gov/cneaf/electricity/st_profiles/toc.html. Federal Reserve Bank of Chicago 5 capability and available transmission capability; market monitoring; planning and expansion; and inter-regional coordination. In 2001, FERC clarified its goals by arguing for the formation of as few as four very large RTOs to cover the entire national grid.4 What does restructuring mean? Electricity is provided to consumers through a very complex mechanism. This mechanism is complex from both a technological and regulatory perspective. On the technology side, providers must match energy supply and highly variable demand by managing different sources of generation that operate at differing levels of efficiency. This process includes taking into consideration scheduled and unscheduled generation shutdowns, changes in fuel prices, seasonal variation, a shifting customer base, and even daily weather. On the regulatory side, electricity policy is the shared responsibility of federal, state, and local policymakers. Jurisdictional boundaries between these various regulators are not often clearly drawn, and policy goals can come into conflict. Given this complexity, it is not surprising that there is no single definition for electricity restructuring. However, in most cases, restructuring focuses on taking the once integrated functions of a traditional regulated utilitygeneration, transmission, and distributionand separating or unbundling them into stand-alone services. In the case of generation, the goal of the unbundling is to introduce competition. In the case of transmission, the restructuring goal is to modernize the transmission infrastructure to support open access to the grid and the most efficient delivery of bulk electricity on both an intra- and inter-state basis. Efficient transmission allows the cheapest power to be used first and reduces the overall peak power or back-up capacity needed in the system. In the case of distribution, it is hoped that unbundling will make it easier to identify the true cost of distributing electricity, thereby eliminating hidden costs and cross-subsidies among end-users of electricity. The starting point for the restructuring debate focuses on creating competition for generation through market deregulation. Vibrant supply competition is at the core of the restructuring argument. On the positive side, choice of generation supply can allow consumers to select more customized electricity service, while putting market pressure on generators to innovate and produce more efficient generation alternatives. Supplying competitive choices for generation would help better manage system peak load demands by providing more options to distribution systems when electricity shortages occur. However, if generation competition fails to develop, eliminating traditional regulatory safeguards can result in 6 consumers being exposed to service provision by an unregulated monopolist. From a practical perspective, promoting competition in generation requires attracting new firms with independent generation sources into the market and encouraging the trading of electricity across the grid. For the most part, electricity generation would eventually be carried out as an unregulated competitive service. Generation would be supplied both by subsidiaries of traditional local utilities and new independent power supply companies that would enter the generation business and sell power into the grid. Also, power marketers (firms that trade electricity) could provide local utilities with contracts and hedges, thereby offering them a wider range of options for managing the energy demands and price risks of customers in their service areas. However, from a theoretical perspective the existence of new suppliers in every market may not be necessary to promote the benefits of opening generation to competition. The threat of competition can provide incentives for existing generators to improve efficiency and offer new products. Still, the high cost of entering the generation business may require the physical presence of a competitor, since existing generators know that a potential rival may have a lag time before it is able to provide new supply into the market. Construction delays, permit requirements, and transmission limitations may affect a competitors ability to offer service. The second unbundled function is transmission. Transmission applies to the bulk movement of power across high voltage power lineslinking individual utilities to sources of power. In the past, integrated utilities tended to favor building limited transmission networks. These networks would often link a single utility with one or two other outlets for importing or exporting power. They were not designed to serve as universal transmission grids for multi-state regions, since most utilities built their own generation systems with large reserves to serve their peak load requirements. Fundamental to restructuring has been the assumption that an independent entity needs to be established to run the transmission system. The grid makes the market and without it, wholesale buyers and sellers will not choose to trade. Without an independent transmission organization (such as an RTO), local utilities cannot be assured of supply and they will still want to establish transmission systems that primarily meet their local needs. If independent generators are unable to access or have uncertain access to the transmission grid, they cannot serve their customers and restructuring is jeopardized. Thus, open 1Q/2002, Economic Perspectives access to a technologically adequate, multi-state transmission system is essential to promoting competitive generation sources. The third unbundled element is the distribution system. While the transmission system serves as the superhighway for moving bulk electricity, the distribution system can be thought of as the off-ramps and local roads that bring electricity into the homes and businesses of consumers. Under restructuring, traditional utilities often create a subsidiary that is purely in the distribution business. Since it makes little sense to build competing distribution networks, these distribution companies are often state-regulated monopolies that not only have responsibility for the wires that run into an individual home or business, but also for billing and other administrative functions. Even with distribution, it is hoped that by unbundling the function, the true cost of operations for specific classes of customers can be more easily identified and priced accordingly. In doing so, distribution operators will focus on efficiency improvements to serve varying classifications of customers. Restructuring at the regional level— The Midwest To date, electricity restructuring continues to be an uneven process. Even states that have expressed similar electricity policy goals, such as improving transmission, reducing noxious air emissions, and attracting new generation, often adopt different strategies. One of the real challenges facing electricity policy in the Midwest is the lack of a consensus on the benefits of opening electricity markets to competition. At first glance, the five states in the Seventh District are extremely heterogeneous in terms of the price of electricity and their interest in pursuing restructuring (see table 3). While Illinois and Michigan continue to press forward with plans to open their electricity markets to competition, Indiana, Iowa, and Wisconsin are taking a decidedly cautious approach. Two areas that the five states can agree on are the need for improvements to the regions transmission grid and the need to account for changes in environmental policy when considering alternatives for future electricity generation. As figure 3 demonstrates, the Midwest is not alone in this piecemeal approach. Throughout the nation, states are choosing different strategies for pursuing restructuring; and the recent problems in California have slowed restructuring activity in several states. On the road to restructuringIllinois and Michigan Illinois Illinois took its first steps on the road to electricity restructuring in 1997 with the passage of the Electric Service Customer Choice and Rate Relief Law. The state is phasing in competition and customer choice. Different classes of customers have been given the option of choosing an electricity supplier, beginning with certain large nonresidential customers in October 1999. As of December 31, 2000, all nonresidential customers can pick a supplier, and a watershed will be reached on May 1, 2002, when retail choice will be open to residential customers.5 However, some analysts suggest that the start of residential choice in May will be met with very little immediate activity. In the service territory of the states largest utilityCommonwealth Edison (Com Ed) residential rate reductions of 20 percent have been ordered. These rate reductions were intended to provide residential customers with benefits of restructuring during the period when nonresidential customers were permitted to choose suppliers. The problem is TABLE 3 Seventh District energy profile Illinois Indiana Iowa Michigan Wisconsin U.S. Exporter Coal Exporter Coal Importer Coal Importer Coal Importer Coal n.a. Coal Average electricity price (cents/kWh) Industrial Commercial 7.46 5.11 7.77 5.34 3.95 6.08 6.04 3.99 6.67 7.09 5.03 7.81 5.44 3.86 5.87 6.74 4.48 7.41 Residential 9.85 7.01 8.38 8.67 7.17 8.26 Exporter or importer Primary generating fuel Note: n.a. indicates not applicable. Source: U.S. Department of Energy, Energy Information Administration 1999, State Electricity Profiles, available on the Internet at www.eia.doe.gov/cneaf/electricity/st_profiles/toc.html. Federal Reserve Bank of Chicago 7 FIGURE 3 Status of restructuring of electricity markets WA ND MT MN ME OR ID WY MI CO MA PA IL UT NH NY IA NE NV VT WI SD IN MO KS RI OH NJ DE WV VA KY CA CT MD DC OK AZ NM TN NC AR SC MS AL GA TX LA Already restructured Ready to restructure FL Investigating restructuring No restructuring plans Source: U.S. Department of Energy, Energy Information Administration, 2000, “The restructuring of the electric power industry: A capsule of issues and events,” report, January. that these rate reductions are likely to discourage new suppliers from entering the market, because they will find it difficult to undercut the price already being offered to residential customers.6 Although Illinois has made progress in the nonresidential market, statewide performance is at best uneven. A number of new service providers have received certification by the Illinois Commerce Commission as alternative retail electric suppliers. These suppliers have been reasonably successful in securing industrial and commercial customers, particularly in the Com Ed service territory (see table 4). For example, the switching rate for the eligible industrial load in Com Eds territory is 72.5 percent; however, the next highest industrial switching rate is only 19.7 percent in Illinois Powers territory. In five of the service territories, no switching has occurred. The switching pattern for eligible commercial customers is similar.7 As is often the case in the opening of a new market, suppliers are largely pursuing the best and most profitable accounts. Whether the advantages of choice are reaching all nonresidential customers remains to be seen. An additional issue is whether these switching rates can be sustained. In Pennsylvania, the states largest utility, Peco Energy, reported losing 44 percent of its industrial customers and 30 percent of its commercial customers to new suppliers when choice was 8 first made available. One year later, Peco had reclaimed many of these customers, leaving their net customer losses at only 4.7 percent for industrial customers and 5 percent for commercial businesses.8 Efforts to encourage competition have occurred not just on the supply side of the equation, but also on the demand side. Buyers have formed collaboratives and secured their own discounts. The most prominent of these represents a coalition of the City of Chicago and 48 suburban governments. This group signed up with Houston-based power marketer Enron to provide their energy needs, and the group estimated that they would save $4 million per year through the new service provider. However, the announced bankruptcy of Enron in December 2001 led to the cancellation of this contract. One final positive development in Illinois has been the states current ability to attract new generation facilities. Illinois is a preferred location for new naturalgas-fired generation, partly due to the presence of major gas pipelines in the state. Currently, 59 plants with a generating capacity of 27,881 megawatts (MW) have either been permitted, are under review, or have been placed in service since 1999.9 While it is unlikely (and not necessarily desirable) to have all of this generation built, clearly Illinois has not faced significant challenges in attracting investment in new plants. These new plants 1Q/2002, Economic Perspectives customers.10 The Michigan Public Service Commission (PSC) has expressed Switching statistics for Illinois concern that a shortage of in-state generation capacity and an inadequate transmisA. Industrial customers sion system are responsible for the lack of Switching rate, Switching rate, response from alternative suppliers. The share of eligible share of all industrial load industrial load PSC reported that as of February 1, 2001, ( - - - - - - - - - - - - - percent - - - - - - - - - - - - - ) only ten alternative energy suppliers had been certified and only four of these were AmerecenCIPS 7.5 6.4 actively serving retail customers. The PSC AmerenUE 0 0 reports that the pilot programs have demCILCO 0 0 ComEd 72.5 39.9 onstrated the importance of transmission Illinois Power 19.7 15.1 in making customer choice effective. WithInterstate Power 0 0 out adequate transmission, new suppliers MidAmerican 4.0 3.6 Mt. Carmel 0 0 are unable to secure and deliver power to South Beloit 0 0 their customers. The existing transmisTotal 38.8 30.0 sion system is physically not adequate to support a vibrant competitive market.11 B. Commercial customers Due to the lack of in-state generation, Switching rate, Switching rate, the PSC reported that Detroit Edison and share of eligible share of all Consumers Energy would need to purchase industrial load industrial load about 2,900 MW (representing roughly ( - - - - - - - - - - - - - percent - - - - - - - - - - - - - ) 15 percent of estimated total demand) over the summer of 2001 to meet load and AmerecenCIPS 30.7 7.1 AmerenUE 0 0 maintain a reasonable operating margin. CILCO 0 0 These structural impediments are limiting ComEd 48 16.5 options even for customers that are activeIllinois Power 11.6 3.0 Interstate Power 0 0 ly interested in receiving service from an MidAmerican 20.2 8.6 alternative provider.12 Mt. Carmel 0 0 Michigans transmission constraint South Beloit 0 0 Total 40.3 12 has been of sufficient concern that the Michigan legislature mandated that utiliNote: Rates are effective through December 15, 2000. ties make provisions for 2,000 MW of inSource: Electric Light & Power, 2001, Illinois and Deregulation: A Fresh Update Places the State at the Halfway Point, March 22. cremental transmission capacity by 2002. As for the shortage of in-state generation, the PSC reports that 2,166 MW of new generation has gone on line since June 1999. have the potential for increasing generation competiGenerators have also reported to the PSC their intention in Illinois; however, in a deregulated system, they tion of adding 7,670 MW in the future, based on fawill not be obligated to serve the Illinois market. cilities that are either planned or under construction. Michigan TABLE 4 Michigan is the other state in the District that continues to press forward on restructuring. Michigan faces a slightly more urgent burden, in that the state is an electricity supply importer. Currently, retail market activity in the state appears less developed than in Illinois. Pilot programs where incumbent utilities have made roughly 10 percent of their load available for customer choice have attracted little active interest. For example, as of January 1, 2001, two large utilitiesDetroit Edison and Consumers Energyhave made a total of 2,100 MW available to retailers to provide to customers. So far, only 257 MW of electricity is actually flowing from alternative service providers to large Federal Reserve Bank of Chicago The cautious approachIndiana, Iowa, and Wisconsin Indiana With low power prices derived from coal-based generation, Indiana has been cautious in pursuing restructuring. Policymakers have focused on maintaining the advantage of low-cost power, and Indiana consumers do not appear to be actively interested in a choice of provider. However, Californias recent bad experience has led Indiana policymakers to continue to study their options. One policy area of interest to the state is the role of merchant power plants (which appear to 9 be interested in establishing locations in the state) and trying to establish a more comprehensive energy policy to guide Indiana decisionmakers. In the last several years, over two dozen merchant plants have been proposed, with the states Public Service Commission approving seven of the plants. Indianas interest in expanding generation may be well founded. The state is currently an exporter of electricity and would like to use its combination of low price and relative energy surplus to attract economic development. However, a recent report of the State Utility Forecasting Group (SUFG) found that it has a declining ability to meet its electricity needs and maintain a 15 percent reserve margin. (Because electricity cannot be stored, reserve margins of 15 percent or greater are considered prudent, particularly when transmission limitations might limit access to generation from more distant utilities.) In its 1999 projection estimates, the SUFG predicts a potential deficit of 2,000 MW by 2005 and of 4,000 MW by 2010 (assuming no new generation is added in state). With neighboring states, particularly Illinois and Ohio, actively adding generation and restructuring to encourage competition, Indiana is anxious to maintain its currently favorable status. Indiana officials are currently reviewing proposals for 2,330 MW of new generation. Iowa In Iowa, the focus of electricity policy has been on incentives for creating new generation, rather than opening markets to competition. A bill considered during the last session of the Iowa legislature offered indirect incentives for new generation by requiring the Iowa Utilities Board (IUB) to specify in advance the rate-making principles it would use for establishing the recovery and return on investment for any new plants built. Also under the proposed legislation, Iowa utilities signing contracts for power from in-state resources would receive irreversible contract approval within 90 days if the IUB found the contract reasonable and prudent. Evidence from the IUBs comprehensive review of Iowas electricity structure and the implications of restructuring suggests that the state needs to focus on generation. The IUB finds that, based on a projection of annual load and annual load obligation, the largest utilities in the state will move from a surplus of 495 MW in 2000 to a potential deficit of 2,208 MW by 2009. The turning point in the surplus/deficit could come as early as 2003.13 Recently, two large utilities have expressed interest in building new generation in the state. MidAmerican Energy announced plans to build a gas turbine plant east of Des Moines that could produce 540 MW of 10 power by 2005. This would be the first new large power plant built in Iowa in 20 years. Another major player, Alliant Energy, is investigating building up to 1,000 MW of new generating capacity in Iowa.14 The provision of electricity in Iowa is relatively complex, with a large number of utilities in the state. Large investor-owned utilities accounted for 76 percent of the megawatt per hour (mWh) sales in 1998, while 137 rural cooperative and municipal companies provided the remainder. Wisconsin Wisconsins electricity policy has focused on improving capacity and reliability. While the states low prices continue to be an advantage, reserves continue to dwindle and transmission bottlenecks have led to concerns about reliability. State policy has not favored sweeping restructuring. Instead, policy has emphasized providing incentives for utilities and independent companies to create new, in-state generation. State electricity supplies are extremely tight, particularly in the eastern half of the state, according to the Wisconsin Public Service Commissions (PSC) Strategic Energy Assessment for 2001. During summer peak generation months, transmission constraints make it difficult to relieve supply shortages in eastern Wisconsin, even when power is available from sources in western Wisconsin and neighboring states. This grid congestion has left major utilities having to rely on interruptible service contracts to prevent outages. Customers subject to these interruptible contracts have become increasingly dissatisfied with this arrangement, even though they receive lower prices in return for permitting the service reduction. By 1998, transmission limitations had reached a point where the PSC recommended adding 3,000 MW of transmission into the state, doubling the existing transmission capacity.15 Generation is also needed. The state has not added any baseload generating units since 1985. Wisconsin has also faced challenges attracting new suppliers. By the end of 1999, the state had only two merchant plants operating; however, the PSC anticipates that merchant plant production could reach 10 percent of the states generating capacity by 2002. In all, merchant plants could add 740 MW of new generation by 2002.16 Wisconsin policymakers have been investigating ways in which the existing regulatory structure can be modified to provide investment incentives while retaining oversight authority. Much like Iowa, Wisconsin is emphasizing adding new rate-regulated generating units and encouraging long-term contracts with in-state independent generators. Proposed policy options include increasing regulatory certainty for recovering new investments, for example by raising the permitted return 1Q/2002, Economic Perspectives on capital investment. These are still proposals, but like Iowa, Wisconsin is trying to chart a path that will increase capacity and transmission quality, without reducing direct oversight of the electricity business. Electricity restructuring at mid-term—What have we learned? Electricity restructuring is a work in progress. New markets and mechanisms will not form overnight and the transition period has already proven to be bumpy. Californias experience with restructuring has led many states to reconsider whether restructuring can produce the benefits of lower consumer prices, more efficient generation, and product innovation often touted by proponents. Given the balkanized nature of the U.S. electricity system, it is understandable that establishing a national electricity policy has proven difficult. However, early programs in the electricity system and experience from the deregulation of other network industries (telecommunications, trucking, airlines, and natural gas) have produced some useful lessons for policymakers to consider. For the purposes of this article, I group the experiences from restructuring into three broad categories. The first category focuses on the unique features of electricity that directly influence its market structures. The uniqueness of electricity limits its treatment as a standard commodity and influences the set of policy goals that can be achieved through restructuring. The second category considers the need to invent or reinvent institutions to govern the industry as it restructures. For existing regulatory bodies, this will mean adding new responsibilities and shedding old authority. Also, entirely new institutions (RTOs in particular) will need to be created and provided with the resources, authority, and mission to manage separated functions such as transmission. The final category addresses issues of market structure and design. This has two components. First, regulatory bodies are facing transition costs as they adapt to dealing with restructuring and the introduction of markets. Second, consumers are facing their own transition costs as they are exposed to a less regulated electricity system. Category 1Understanding electricity as a unique commodity In some important ways, electricity is unlike other commodities. It is a modern necessity; we rely on it to provide light and heat, to fuel commercial and industrial production, and to run most appliances in the home. Much like water, electricity generally carries a low price, has a high value to the consumer, and offers no short-run substitutes. The critical role of electricity has led to a regulatory policy favoring the development Federal Reserve Bank of Chicago of excess capacity to ensure reliability of service, even when this has had the effect of inflating the price. Second, electricity has certain physical properties that make it different from other commodities. Primarily these have to do with difficulties in storing or rapidly adding new capacity. Significant supplies of electricity are almost impossible to store and electricity needs to be consumed when produced. This characteristic, combined with the inelastic demand of customers, creates high marginal prices when there is a shortage of electricity in a market, unless power can be imported through the grid. Assuming that a utility has no other choice but to meet its load obligation through the spot wholesale market, it is fully exposed to paying whatever price the market will bear for a commodity that is consumed immediately. In addition, it is difficult to create new electricity capacity quickly. New power plants in most states require 18 months to 24 months to site and build, and local opposition to construction often lengthens the process even further. Moreover, new power plants require very high capital outlays and can be seen as risky investments in the restructured electricity market. The decision to build a plant is predicated on the cost of key variable operating costs (fuel in particular) and an assessment of the price that can be charged for electricity, which is often dependent on regulatory decisions. New generators are often providing reserve, back-up power, rather than meeting the daily baseload needs of a given service area. This also introduces uncertainty into the decision to add new generation. These factorsthe inability to economically store electricity and the inability to create new generation quicklycreate the conditions for high prices in the spot wholesale market and can provide certain sellers of electricity with the opportunity to charge very high prices for power. To avoid very high transition costs associated with restructuring, state policymakers need to ensure either that adequate reserves exist or that restructuring will create the conditions for adequate reserves. Another special feature of electricity is that it needs to be delivered over an extensive physical grid. The transmission grid is the physical market for trading electricity, making issues of grid reliability, capacity, and access critical in creating the conditions for competitive market operations. Assessing this infrastructure requires integrating several levels of analysis. First, what is the current condition of transmission and generating assets? Can steps be taken to improve the operation of the existing facilities? How easy is it to access the grid and efficiently transport electricity to where it is needed? Recently, electric power forecasters 11 have questioned as inadequate the amount of new transmission capacity that is planned in the system. Second, will new technologies change the need for the type of generation and transmission that is needed? Some analysts suggest that micro-generation, fuel cells, and other new technologies may allow an increasing number of consumers to generate electricity in their homes and businesses. If this is true, building large generation plants and making extensive upgrades to the grid may be less important than has traditionally been assumed. The performance of the transmission grid (as we have seen in the survey of the Seventh District states) can undermine the development of healthy wholesale markets and limit the appeal for new companies to develop new operations in Midwest states. Blackouts in Chicago and power problems in eastern Wisconsin have frequently been linked to transmission bottlenecks rather than a lack of adequate generation. The policy problem is that many decisions regarding transmission and the construction of new generation need to be conducted in coordination with actions in neighboring states. Improving the intrastate grid is certainly important, but it will not fully substitute for the need to carry out such a policy in concert with activities in neighboring states. A well-functioning grid can reduce the need for building excess generation capacity, if energy can reliably be made available through transmission. For example, in Wisconsins forecast of energy needs, the Public Service Commission makes differing recommendations for generation needs and reasonable generator reserve requirements based on assumptions about the future performance of the grid. With a well-functioning grid, the traditional reserve requirements of 15 percent are adequate, but without any improvements in grid performance, recommended safe generation margins for utilities facing poor grid connections rise to 30 percent.17 Based on this assessment of the infrastructure for delivering electricity and the conditions that make electricity a special commodity, policymakers need to have a clear set of restructuring goals. Artificially reducing the retail price of electricity in order to gain support from various constituent groups should not be the sole motivation for restructuring. The price of electricity should reflect market factors. If competition introduces efficiency that lowers prices while maintaining other important policy goals, such as reliability and adequate reserves, lower prices can be a welcome outcome. In testimony before the Senate General Administration Committee, economist Paul Joskow suggested that deregulation is not a goal in and of itself. 12 The goal is to create well functioning competitive markets that perform better than the regulated structures they replace.18 The real benefits of restructuring will become apparent when fully functional markets are operating. This will take time. Frequently, electric restructuring is presented as a policy designed primarily to reduce the current price paid for electricity. Often this is motivated by states facing the highest electricity prices being the most interested in opening their electricity market to competition. It is no accident that in the Seventh District, Illinois and Michigan, the high-price states, are pursuing restructuring, while the three low-price states are hesitating. This interest in promoting competition has been pressed by large industrial customers that would expect to pay significantly less for power in an open wholesale market with multiple suppliers bidding for their business. Under fixed regulated retail tariffs, industrial customers often argue that they are subsidizing the residential market. However, an immediate policy objective of reducing the price paid for electricity may impede valuable longer-term policy goals, such as encouraging innovation and efficiency in generation and promoting product customization. For example, many high-tech firms are more interested in electricity that can be provided with complete redundancy (that is, 100 percent back-up capability) and is of the highest quality, not subject to transmission or distribution disruptions. Clearly, the opportunity to purchase this type of specialized, premium product may be of greater interest to certain classes of customers. Similarly, some consumers may want to promote environmental goals and may be willing to pay a premium for electricity from renewable sources. For example, in the Pacific Northwest, the Bonneville Environmental Foundation and the Climate Trust of Portland are marketing green tags to utility customers. These tags sell for $20 and the proceeds are used to purchase power from a renewable source such as wind power, thereby allowing individual customers to purchase offsets to traditional energy sources.19 Beyond the often-repeated goal of reducing prices, longer-term goals of establishing competitive markets in electricity should include establishing incentives for investments in more efficient power plants, stimulating the introduction of new technologies (such as fuel cells), encouraging innovation in both supply and demand-side management, and even providing incentives to promote needed investments in transmission and environmental mitigation. Policymakers clearly face some political challenges as well, which may lead some states to adopt shortrun policies (such as mandatory price caps and price 1Q/2002, Economic Perspectives cuts for residential customers) that would in fact prevent changes in wholesale markets from being reflected in retail markets. The tradeoffs between lower prices, increased investment, and even related environmental goals of lower emissions will be made in a political context. Ideally, the political process will set the policy targets and leave the methods for achieving these targets efficiently to market mechanisms. It is also important that the institutions charged with implementing and overseeing the restructuring process be insulated from political interference. Ultimately, this requires an electricity policy characterized by integrated resource planning, a complete understanding of technology and efficiency tradeoffs, and the flexibility to make mid-course corrections as changes in supply and demand conditions require. Category 2Inventing or reinventing institutions and roles We know by now that electricity restructuring is a complicated business. Existing institutions such as the state public utility commissions and FERC will have new roles in guiding restructuring and they will be asked to reduce their authority in areas where they have traditionally held jurisdiction. Even more difficult will be creating new institutions, the proposed RTOs, market surveillance committees, and electricity trading systems to support restructuring. Both the traditional regulators and these new institutions need to be vested with the resources, incentives, and authority to carry out their missions. They need the resources to actively monitor the markets under their authority and the data to know what is driving these markets. Governance issues include clarifying overlapping jurisdiction and decisionmaking authority. Resource issues include staffing, staff training, and sufficient data to monitor an industry that is undergoing profound changes. Californias experience again presents a telling lesson. For example, Governor Gray Davis acknowledged that when it came to negotiating for power sales to the state, the states negotiating team faced a circumstance similar to a tee-ball team playing the NY Yankees.20 While this was an extreme situation, it underscores the difficulties that may arise when the state government is forced to play an unfamiliar role. To date, the formation or reinvention of these institutions has been a difficult process; and the related uncertainty may be discouraging potential infrastructure investments. The most obvious example in the Midwest has been the attempt to form an RTO. Most of the regions major utilities had joined the Midwest Independent Systems Operator (MISO) and assumed that this organization would become the RTO for the region. However, internal issues led to many of the Federal Reserve Bank of Chicago largest utilities withdrawing from MISO and establishing the Alliance RTO. This group subsequently identified the British firm, National Grid, as its grid operator. However, it is unclear whether National Grid would actually own any of the transmission infrastructure or would simply serve as the system operator. If the traditional utilities maintain their ownership of their existing transmission assets, it is unclear how investment decisions will be made or coordinated. These developments also call into question how easy it will be for competitive generators to access the grid on equal terms. Adding to this confusion is FERCs recent request that the geographic boundaries of the RTOs be expanded in an effort to develop a national grid. FERC has suggested that MISO, Alliance, and the Southwest Power Pool should consider combining into a superregional RTO.21 (On December 19, 2001, FERC approved MISO as the RTO for the Midwest. The MISO transmission area will operate in 20 states. In its action, FERC suggested that the Alliance organization could operate as a member of MISO.) Even if the issue of what organization is running the RTO is resolved, the potential functions of the organization need to be considered. One white paper produced by the Electricity Policy Research Institute examining power issues on the West Coast produced a list of recommendations for improving electricity operations. The study suggests that RTOs should have either primary or shared responsibility for addressing a significant array of issues, including repairing dysfunctional wholesale markets; generating standardized regional energy information; and implementing a whole system reliability-centered maintenance capability. This whole-system approach would include assessing the equipment health for vital components; initiating comprehensive, region-wide transmission risk analysis; creating a seamless real-time exchange of information among regions; coordinating training of grid operating personnel; developing power-flow technology for system reservation and scheduling; and establishing regional authority for siting and cost sharing.22 Clearly, this is an ambitious agenda for an organization that is still being developed. Finally, the formation of large, multi-state RTOs suggests that FERC, and not the state public utility commissions, will need to play the major regulatory role over transmission issues. Even this decision is not clear-cut. Following the summer 2001 meeting of the National Governors Association, the governors agreed to work with the U.S. Department of Energy and other federal agencies to improve transmission, but made it clear that the states still wanted to maintain their traditional policymaking role over 13 transmission. Specifically, the governors issued a statement to the effect that governors oppose preemption of traditional state and local authority over siting of electricity transmission networks, but governors recognize that situations exist where better cooperation could improve competition and reliability. Governors are willing to engage in a dialogue with the federal government and industry to address these situations in a manner that does not intrude upon traditional state and local authority.23 A final governance issue has to do with the oversight of municipal and cooperative utilities. These utilities, which are not major suppliers of electricity in most states, are often self-governed. While larger investor-owned utilities are subject to regulation by state utility commissions and FERC, it is unclear whether these smaller players should be brought under the same regulatory structure. Category 3Market structure and design Transition costs for regulators Industry observers assume that the introduction of competition into the generation component of the electricity system will best be accomplished through market mechanisms. The eventual goal of promoting unregulated (or at least minimally regulated) competitive generation is to promote cost efficiency and greater diversity of generation resources. Yet, the introduction of markets provides another set of challenges. These challenges can be understood across three dimensions. The first dimension relates the uniqueness of electricity as a commodity to the implications of using markets to deliver electricity, as discussed earlier. The second two dimensions consider two sets of transitions costs of moving to a market structure. The first set of costs relates to the actions and adaptation (and potential rigidity) of regulatory bodies and market participants in responding to the unbundling of electricity service. The second set of transition costs relates to the implications of market structures for end-users/consumers of electricity. I discussed earlier how the physical properties of electricity create certain challenges to establishing smooth operating markets. The inability to store or rapidly create new capacity, as well as physical limitations of the grid as the market trading system are all factors that must be accounted for in successfully restructuring electricity. Another important change that is brought about by moving to a market structure is the potential lack of incentive to provide large generation reserve margins. In the old regulated system, the utility would be willing to build surplus capacity into its generation plans. Since the regulator permitted 14 a rate structure that allowed the utility to recover the cost of this extra capacity (even if it went unused), there was little risk involved in carrying a large reserve margin. In moving to an unregulated market structure for generation, carrying reserve capacity (particularly when the generator cannot store the power) clearly makes little sense from the generators perspective. It is therefore not surprising that states that have been moving toward restructuring have seen reserve margins decline, so that supply and demand more closely match each other. Another consequence of moving to a more market-based electric system is that the new market structure may provide opportunities and incentives for suppliers to exercise market power. Market power can be understood as the ability to raise market prices through unilateral action so as to profit from the price increase. Concern over suppliers exercising market power is one of the potential transition costs faced by regulators. To address these concerns in the wholesale market, FERC only allows suppliers that can demonstrate that they do not have market power to sell in the market and receive the market-clearing price. Suppliers that cannot demonstrate an absence of market power are limited to charging the cost-ofservice rate set by FERC.24 Establishing whether a firm has market power is critical in determining whether prices being charged in newly created electricity markets are the product of manipulation or genuinely reflect the interaction of supply and demand. In the case of the California market experience of recent years, it was often alleged that certain suppliers would withhold generation or use other timing and bidding techniques to receive extraordinary prices when it was known that utilities would have to make purchases in the spot market. Determining the prospective market power of an electricity supplier is a very difficult business. Attempts by FERC to define market power have been met by skepticism. Frank Wolak, a Stanford University economist who serves as chair of the Market Surveillance Committee of the California Independent System Operator, points out that market power is often incorrectly estimated based on the concentration indexes applied to geographic markets. These geographic boundaries fail to account for the fact that electricity must be provided to final customers over the existing transmission grid. Limitations in the grid can make differences in the bidding, scheduling, and operating protocols of the market crucial in determining whether a supplier can exercise market power. Work by Wolak, Borenstein, and Bushnell (2000)25 measured the extent of market power in California since 1998, 1Q/2002, Economic Perspectives and the Market Surveillance Committee has contributed a number of reports on the subject. By the summer of 2000, the committee found that average monthly prices being charged for June were 182 percent above what would have been expected if no generator was able to exercise unilateral market power.26 Competitive generation and supply will change the incentives and behavior of many firms in the electric supply business. In the case of traditional integrated utilities, the spinning off of generation into unregulated affiliated companies raises the expectations for shareholders that these generation companies can become profit centers for the parent company. On the one hand, increasing the importance of profitability as a measure of success for the generation company should promote efficiency. On the other hand, it also raises the incentive for generators to take advantage of market conditions to receive the highest price for their production. It is important to recognize that the public service ideal that once guided utilities will be de-emphasized once generation is treated as a commodity. Studies by Californias Independent Systems Operator provide evidence that generators did withhold supply in order to bid up market prices. During the fall and winter of 200001, there were nearly four times as many scheduled and unscheduled plant shutdowns as in the previous year. While some of these could be attributed to breakdowns in older plants that were forced to run at higher capacities than intended in order to avoid blackouts and brownouts, some shutdowns seemed to be more strategic. This motivation is even more obvious in the case of independent merchant generators. An example of this occurred in the California market in January 2001. In the latter half of the month the California grid operator faced a series of conditions that made the likelihood of an energy shortage in the state highly probable. A combination of unfavorable weather, a lack of supply from traditional reserve supplies from northwestern states, and a malfunction at a 1,000 MW plant in the state meant that the grid operator had to scramble to find power. Eventually, the operator found a California merchant plant that was willing to offer power, but only at the record price of $3,880 per mWh. The grid operator ended up buying the power, and this situation demonstrated the ability of this single supplier to set the price in the market. In justifying the record price charged, the merchant generator admitted that the price was less related to the plants cost of generating power than the risk premium it was charging for selling into the financially shaky California energy market. (As it turned out, the generator never received the $3,880 per mWh. First, FERC investigations into Federal Reserve Bank of Chicago the price charged found that the generator had overcharged for the electricity and reduced the price to $273 per mWh. In fact, to date the generator has only received payments equaling $70.22 per mWh, due to the inability of the purchasers to make good on their debts.)27 The Independent Systems Operators study also demonstrated how independent generators could use bidding to influence the price in the wholesale market. It was estimated that the combination of withholding supply and strategic bidding behavior accounted for one-third to one-half of the increase in prices in the California wholesale energy market. This added roughly $6 billion to the costs California consumers had to pay. In the case of bidding behavior, the study revealed that the rules of the hourly auctions provided an opportunity for generators to manipulate prices to their advantage. Under the terms of the hourly auction, generators would offer batches of energy at various prices. The system operator would then rank the bids according to price, and the price in the market was set at the bid price for the last unit of electricity needed to meet the demand on the grid. At this point, all generators in the auction would receive the price that had been paid for the last unit. The concept behind this bidding structure was to treat the electricity supply like any other commodity, where everyone in the market would receive roughly the same price for providing a homogenous good. Over time, generators became very savvy about taking advantage of this structure when it seemed likely that supply would run short. Essentially, most bids would be offered at prices that roughly reflected the cost of generation plus some reasonable margin. However, the final units would be bid at an extreme premium, sometimes at ten times what the normal price would be. If all of the bids for these final units of supply came in at these prices, the operator had no choice but to accept this price in order to meet load. At this point, because of the terms of the market, all of the supply bid that day would receive this high price. While it has not been shown that this bidding behavior involved collusion among the generators, it is clear that this auction system provided an opportunity for savvy bidders to take advantage of the process, and it appears that they did.28 Critics of Californias restructuring plan have suggested that permitting the use of long-term contracts and other hedges would have made it is far less likely that the prices bid would have been at such exaggerated levels, since spot shortages would have been less frequent. Finally, the role of power traders in markets needs to be understood. These firms provide financial options to electricity markets, but are not in the 15 business of building or owning generation facilities. Firms such as Dynegy are well known as power traders but, increasingly, regulated utilities carry out trading activities through unregulated subsidiaries of their holding company. In an ideal world, this can help utilities and customers manage risk. For policymakers, marketers are often a new institution to deal with. Utility commissions often lack the staff to monitor the behavior of traders in the electricity market for signs of collusion or unfair practices. Policymakers need to understand that the purpose of these firms is to make money through trading, not to serve as a public utility. In this regard, oversight of these firms may best be accomplished through the same mechanisms that govern other commodity trading operations. However, most states appear to lack such a structure or a real understanding of how to deal with electricity traders. In the case of California, for example, market problems identified by the states Market Surveillance Committee were largely ignored. Transition costs for consumers What are the implications of electricity restructuring for consumers? First, consumers need to understand that opening markets will expose them to both the advantages and disadvantages of market pricing. In the past two years, this increasingly has meant dealing with a commodity with high price volatility. Electricity consumers are not used to dealing with market risk, since the regulated electricity system had firm tariff-based prices. Exposing consumers to market-based (or even real-time) prices is a natural consequence of moving to a market system. However, in many cases, states are protecting their residential and small business customers from price volatility by freezing electricity prices for a transition period. This presents several problems. Frozen retail rates mean that consumers are not exposed to the underlying dynamics that are being reflected in the deregulated wholesale market. Californias experience has shown the problems that can result from this approach. Because Californias consumers were insulated from market risk and volatile prices, they never received the appropriate price signal that would have caused them to immediately reduce consumption when electricity prices spiked. This eventually led to the financial insolvency of the states two largest investor-owned utilities. Eventually, Pacific Gas and Electric and Southern California Edison ran up $9 billion in debt, purchasing power in the spot market.29 Only much later did the California Public Utility Commission allow these two utilities to charge higher prices for electricity and, by then, Pacific Gas and Electric Company had filed for bankruptcy protection 16 and Southern California Edison had gone to the state legislature asking for assistance. Efforts to set prices for certain classes of customers during a transition period have other shortcomings as well. If the price is set artificially low, new entrants to this competitive market will not appear, since the margin will not be sufficient for them to capture customers. This will undermine the development of a competitive market. On the other hand, if the price is set too high, consumers may be paying too much in return for stable electricity prices. Instead, it would make more sense to allow prices to reflect market fundamentals. The downside is the resulting price volatility faced by end-users. In order to protect riskaverse consumers from being fully exposed to price swings while these markets develop, risk management tools (hedges and long-term contracts) can be used. Consumers who have grown accustomed to a firm price for their electricity bills can still be provided with this option in a deregulated market. The distributing utility can offer a customized product that protects the consumer against volatility by offering a firm price that has an insurance premium built into the rate. Even now, many utilities offer customers fixed monthly payments that protect them from high electricity bills caused by seasonal conditions. Volatile prices can be an essential element in encouraging more efficient demand-side management. Pilot programs in real-time pricing demonstrate that consumers will respond to price spikes by reducing consumption.30 In testimony before a Senate panel, Joskow went as far as to suggest that the default service option for larger commercial and industrial consumers should be purchasing electricity at real-time prices. He argued that the use of real-time pricing for these more sophisticated customers would introduce demand elasticity into the wholesale market and this, in turn, would dampen price volatility and help mitigate supplier market power.31 Providing more opportunities to manage peak load needs can produce a more efficient electricity system. Allowing price signals to be felt can be an important motivator in improving demand-side management programs. Another transitional cost to consumers is electricity reliability. In the bundled service, rate-regulated, historical model for providing power to the consumer, the blended tariff rate ensured that investments would occur in all aspects of electricity provisionincluding customer service and reliability. Once the service is separated into three components, the low-profit regulated portions of the business (distribution in particular) may not attract needed investment, which may impair reliability and even service quality. This has 1Q/2002, Economic Perspectives been a frequent complaint in telecommunications restructuring, where once-regulated local phone companies have been allowed into open markets. Once in these markets, the company pursues the most profitable segments of the business, often to the detriment of investing in basic service. Conclusion Electricity restructuring is at a crossroads. Experience to date has brought into focus the difficulties involved in restructuring the industry efficiently through a combination of regulated and deregulated structures. The recent experience of California and the number of issues complicating this transition would be of less concern if it werent for the fundamental role that electricity plays in supporting modern society. There are still good reasons to believe that electricity restructuring can fulfill its early promise. However, as California has demonstrated, electricity restructuring must be fully thought through and carefully crafted. Policy missteps can lead to unintended costs that will be borne well into the future. At a minimum, electricity restructuring requires a clear set of policy targets that establish goals for system efficiency, investment, and prices. Once these goals are established, institutions must be designed and equipped to meet them and, importantly, must be protected from political interference while they pursue these objectives. NOTES This policy preference favoring geographically defined, integrated utilities was established in the Public Utility Holding Company Act of 1935. 16 Wisconsin Public Service Commission (2000), p. 2. 17 Ibid., p. 4. U.S. Department of Energy, Energy Information Administration (2000), p. 16. 18 Joskow (2001), p. 8. 19 Knight-Ridder Tribune Business News (2001). 20 Smith and Emshwiller (2001), p. 5. 1 2 U.S. Department of Energy, Federal Energy Regulatory Commission (1999). 3 U.S. Department of Energy, Federal Energy Regulatory Commission (2001). 4 U.S. Department of Energy, Federal Energy Regulatory Commission (2001). 21 5 Illinois Commerce Commission (1999), p. 2. 22 Energy Policy Research Institute (2001a), pp. 56. 6 Cvengros (2001), updated August 1. 23 National Governors Association (2001), p. 5. 7 Davis (2001), Illinois page. 24 Wolak (2001), p. 4. 8 OGrady (2001). 25 Wolak, Borenstein, and Bushnell (2000). 9 Cvengros, op.cit. 26 California Independent System Operator (2000). Ibid. 27 www.washingtonpost.com, August 18, 2001. Michigan Public Service Commission (2001), p. 5. 28 Pearlstein (2001), p. A9. 12 Ibid. 29 Ibid. 13 Iowa Utilities Board (2000). 30 Energy Policy Research Institute (2001b). 14 DeWitte (2001). 31 Joskow (2001), p. 14. 15 Wisconsin Public Service Commission (1998), p. 2. 10 11 Federal Reserve Bank of Chicago 17 REFERENCES California Independent Systems Operator, 2000, An analysis of the June 2000 price spikes in the California ISOs energy and ancillary services market, report, available from the Internet at www.caiso.com/docs/09003a6080/07/dc/ 09003a608007dc78.pdf, September 6; other reports on market power and market behavior are also available at the website. National Governors Association, 2001, Comprehensive National Energy Policy, point 18.5, Improving Energy Transmission, Washington, DC, policy position, No. NR-18, p. 5. Cvengros, Laura, 2001, Restructuring activities by state, Indiana Utility Regulatory Commission, report, available on the Internet at www.ai.org/iurc/ energy/restruct/restruct_index.html, updated August 1. Pearlstein, Steven, 2001, The $3880 megawatthour; How supply, demand, and maybe market power inflated a $273 commodity, Washington Post, August 21, p. A9. Davis, Kathleen, 2001, Know your power: EL&Ps state of deregulation features delve into individual battles with electric restructuring, Electric Light & Power, available on the Internet at http://elp.pennnet.com/, Illinois page. Smith, Rebecca, and John R. Emshwiller, 2001, Hurt by deregulation of utilities, California gives itself lead role, Wall Street Journal, available to Internet subscribers at www.wsj.com, July 17, p. 5. DeWitte, Dave, 2001, MidAmerican Energy plans two plants, rate freeze in Iowa, Knight Ridder Tribune Business News: The Gazette, Cedar Rapids, IA, July 11. OGrady, Eileen, 2001, On second thought, Wall Street Journal, available to Internet subscribers at www.wsj.com, September 17. U.S. Department of Energy, Energy Information Administration, 2000, The restructuring of the electric power industry: A capsule of issues and events, paper, No. DOE/EIA-X037, January, p. 16. Energy Policy Research Institute, 2001a, Western States power crisis, June 25, pp. 56. U.S. Department of Energy, Federal Energy Regulatory Commission, 2001, Order initiating mediation, docket, No. RT01-100-000, July 12. , 2001b, Real-time pricing could signal consumer conservation, EPRI Journal online, available on the Internet at www.epri.com/Palo Alto, CA. , 1999, FERC order 2000, docket No. RM99-2-000, order, No. 89FERC 61,285, December 20. Illinois Commerce Commission, 1999, A consumers guide to electric service restructuring, available on the Internet at www.icc.state.il.us/pluginillinois, September, p. 2. Wisconsin Public Service Commission, 2000, Strategic energy assessment, docket, No. 05-ES100, December 18, p. 2. Iowa Utilities Board, 2000, Facts concerning the consumption and production of electric power in Iowa, Des Moines, Iowa, August. Joskow, Paul L., 2001, Statement of Professor Paul L. Joskow before the Senate Committee on Governmental Affairs, U.S. Senate, June 13, p. 14. Knight-Ridder Tribune Business News, 2001, Green tags finance renewables to cut carbon emissions, September 12. , 1998, Report to the Wisconsin legislature on the regional electric transmission system, Madison, WI, September 1, p. 2. Wolak, Frank A., 2001, Statement of Frank A. Wolak before the Senate Committee on Governmental Affairs, U.S. Senate, Washington, DC, June 13. Wolak, Frank A., Severin Borenstein, and James Bushnell, 2000, Measuring market power in the California electricity market, mimeo, August. Michigan Public Service Commission, 2001, Status of electric competition, Lansing, MI, February. 18 1Q/2002, Economic Perspectives The aggregate effects of advance notice requirements Marcelo Veracierto Introduction and summary It is well known that the performance of labor markets, measured in terms of unemployment rates or employment to population ratios, is much stronger in the U.S. than in many European countries. In order to improve the performance of European labor markets then, it is important to determine the cause of these differences. While the degree of unionization, the unemployment insurance system, or minimum wage legislation can have significant effects, most of the literature has focused on firing restrictions as the main candidate. Firing restrictions stand out because they are relatively minor in the U.S., compared with many countries with poor labor market performance. Firing restrictions take several forms in these countries. The most common forms are severance payments, advance notice requirements, and procedural constraints. Severance payments are mandated payments that the employer must give to the worker at the time of employment termination. They vary as a function of the years of service and the perceived fairness of the dismissal. Advance notice requirements impose a pre-notification period that delays the time of employment termination. In turn, the procedural constraints require employers to seek authorization from an outside party prior to performing a dismissal (the outside party being a union, a work council, the government, or the courts). Usually, the authorization procedure is long and costly, and the employer is forced to provide full pay to the worker while the procedure is underway.1 The theoretical literature has typically modeled these forms of firing restrictions in a very simple way: as firing costs that involve either a fixed loss of resources or a fixed payment to the government per unit reduction in employment (firing taxes). While this may be a good first approximation, not many attempts have subsequently been made to model more explicitly the different forms of firing restrictions. The purpose of Federal Reserve Bank of Chicago this article is to analyze the effects on aggregate output, wages, employment, and welfare levels of one particular form of firing restriction, namely, advance notice requirements. I also provide a comparison with the effects of firing taxes to assess the differences between both types of policies. The empirical literature provides good reasons to analyze advance notice requirements separately from other forms of firing restrictions: It suggests that they may have different effects. In a very influential paper, Lazear (1990) constructed two measures of job protection for a set of 22 countries: the amount of severance payments that employers are required by law to pay to blue-collar workers with ten years of experience at the time of termination; and the period of advance notice that employers are required to give to this same class of workers. Lazear then compared these measures with measures of labor market performance, such as employment to population ratios and unemployment rates. Table 1 (overleaf) reproduces the average employmentpopulation ratios, severance payments, and advance notice requirements between 1956 and 1984 for the 22 countries. Figure 1 plots the average employmentpopulation ratios and severance payments, showing a negative relation between the two variables. However, this analysis does not take into account the large variations in labor market institutions over time within each of these countries: Generally, job security provisions were introduced in the 1960s, reinforced in the 1970s, and somewhat loosened in the 1980s. To account for this time variation, Lazear performed a panel data analysis, using yearly observations for each country to regress severance payments against employmentpopulation ratios. His results indicate Marcelo Veracierto is a senior economist at the Federal Reserve Bank of Chicago. The author thanks seminar participants at the Chicago Fed for their comments. 19 payments but opposite results for advance notice requirements. Indeed, they Data for sample countries, 195684 found that longer notice intervals are asEmployment Advance sociated with statistically significant in/population Severance pay notice creases in employment and labor force (months of wages) (months) participation rates. In the theoretical literature, an early Austria 0.43 0.93 3.00 Australia 0.41 0.00 0.00 study of the effects of firing costs was Belgium 0.38 1.24 1.00 provided by Bentolila and Bertola (1990). Canada 0.38 n.a. n.a. Taking factor prices as being exogenous Denmark 0.46 0.48 6.00 Finland 0.47 n.a. n.a. to their analysis (that is, using a partial France 0.40 5.24 1.86 equilibrium setting), Bentolila and Bertola Germany 0.43 1.00 1.86 studied the consequences of imposing Greece 0.37 1.00 10.00 Ireland 0.35 0.00 0.00 firing costs on a monopolist facing a Israel 0.33 8.41 n.a. shifting demand for its product. In that Italy 0.37 15.86 n.a. context, firing costs potentially have two Japan 0.48 0.00 n.a. Netherlands 0.35 n.a. 2.00 opposing effects. On one hand, firing costs Norway 0.42 12.00 3.00 induce the monopolist to avoid large conNew Zealand 0.38 0.00 n.a. tractions in employment after reductions Portugal 0.37 3.36 2.59 Spain 0.35 13.56 n.a. in demand in the hope that demand will Switzerland 0.49 0.00 1.00 increase in the near future. On the other Sweden 0.48 0.00 0.76 hand, they make the monopolist less willUnited Kingdom 0.44 n.a. 0.90 United States 0.39 0.00 0.00 ing to hire workers after increases in demand because of the prospective firing Note: n.a. indicates not available. costs that will have to be paid when deSource: Lazear (1990). mand shifts down in the future. Under a parameterization that reproduces that introducing severance payments of three months observations from European countries, Bentolila of wages is typically accompanied by a decrease in and Bertola found that the first effect is the most the employmentpopulation ratio of about 1 percent. FIGURE 1 Figure 2 plots average employment Severance payments population ratios and average advance severance notice requirements. The plot shows a 18 negative relation, but one that is much weaker than that in the previous figure. 16 However, when the time variation within 14 countries is taken into account, Lazear found that advance notice requirements 12 reduce employment even more than sever10 ance payments. He considered this result to be surprising: At worst, the employer 8 could treat notice requirements as sever6 ance pay, simply by telling the worker not to report during the notice period and pay4 ing him anyway (Lazear, 1990, p. 712). 2 In a later paper, Addison and Grosso (1995) provided revised estimates for the 0 effects of severance payments and advance -2 notice requirements. Including some ad0.30 0.32 0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50 ditional countries and correcting some data employment/population errors from Lazears study, Addison and Source: Lazear (1990). Grosso found similar effects for severance TABLE 1 20 1Q/2002, Economic Perspectives component, since employers must pay wages during important: Firing costs actually increase the average the notice period, despite not needing the workers. employment level of the monopolist. But advance notice requirements have an additional Hopenhayn and Rogerson (1993) performed a effect: They hold workers to their jobs during the pericher analysis, which allowed factor prices to clear riod of notice. This is an important effect. I find that, markets instead of treating them as exogenous (that contrary to firing taxes, advance notice requirements is, they performed a general equilibrium analysis). do not have a negative effect on aggregate employIn their model economy, production is carried out by ment. However, the welfare costs of advance notice a large number of establishments that are subject to requirements can be substantially larger. changes in their individual productivity levels, which To gain a better understanding of these results, I induce them to expand and contract employment over start with a partial equilibrium version of the model, time. Households supply labor, own the establishments, in which prices are fixed. Comparing the effects of adand have access to perfect insurance markets. In that vance notice requirements in this setting with those framework, Hopenhayn and Rogerson introduced fircorresponding to the general equilibrium framework ing taxes that were rebated to households as lump sum allows me to isolate the importance of equilibrium transfers. In this model, firing taxes give rise to an imprice changes to the results. I also analyze the effects portant misallocation of resources. The reason is that of advance notice requirements assuming that once a establishments that switch to a low individual producworker is given advance notice, his productivity on tivity level do not contract their employment as much the job decreases quite substantially. This version of as they should in order to avoid current firing taxes. the model not only adds some realism, but also shows On the other hand, establishments that experience that shirking behavior is an important variable to conhigh individual productivity levels do not expand sider when analyzing advance notice requirements. their employment enough, because they try to avoid The article is organized as follows. In the next paying firing taxes in the future. This misallocation section, I analyze the effects of advance notice in a of resources across establishments reduces labor propartial equilibrium framework. Then, I study the efductivity quite substantially. The decrease in labor fects in a general equilibrium model. Next, I incorpoproductivity induces a large substitution from market rate the assumption that advance notice requirements activities toward leisure and leads to a reduction in generate shirking behavior by workers. Finally, I comtotal employment. This effect can be quite significant: pare advance notice requirements with firing taxes. Firing taxes equal to one year of wages reduce employment by 2.5 percent. Hopenhayn and Rogerson (1993) also calculated the welFIGURE 2 fare costs associated with the firing taxes. Advance notice They found that a permanent increase in advance consumption of 2.8 percent is needed to 12 leave agents in the equilibrium with firing taxes indifferent with moving to the equi10 librium without firing taxes. The model I use in this article is sim8 ilar to that analyzed by Hopenhayn and Rogerson (1993). The main difference is that I use it to evaluate not only the effects 6 of firing taxes in general, but also the particular effects of advance notice require4 ments. The model introduces advance notice requirements in a very parsimonious way. 2 If an establishment decides not to give advance notice to any of its workers, the fol0 lowing period it can expand its employment but not reduce it. On the other hand, if an -2 establishment gives advance notice to some 0.30 0.32 0.34 0.36 0.38 0.40 0.42 0.44 0.46 0.48 0.50 of its workers, it cannot rehire them duremployment/population ing the following period.2 Clearly, advance Source: Lazear (1990). notice requirements have a firing penalty Federal Reserve Bank of Chicago 21 Partial equilibrium In this section, I analyze the effects of advance notice requirements assuming that prices are fixed. The purpose is to isolate the partial equilibrium effects of advance notice requirements from those that arise from general equilibrium effects (that is, from equilibrium price changes). Consider the problem of a single producer facing a constant wage rate w and interest rate i. For simplicity, I assume that output depends only on labor input. In particular, the production function is given by: yt st ntH , where yt is output, nt is the labor input, g is a parameter governing the productivity of labor, with 0 < g < 1, and st is a productivity shock (that is, a higher value of st means that more output can be produced with the same labor input). The shock st follows a Markov process with transition matrix Q. That is, Q(s,s¢) is the probability that st+1 = s¢, conditional on st = s. When no government regulations are introduced, the establishment chooses its labor input to equate the marginal productivity of labor to the wage rate, that is,3 1) w st H ntH 1 . However, I am interested in studying how the behavior of the establishment is affected by the introduction of advance notice requirements. In principle, the establishment could give firing notice to all its workers every period and rehire them at will in the following period, according to the value that the productivity shock takes. If this were possible, advance notice requirements would clearly have no effects. But, given that this alternative is not available to employers in the real world, I consider a notice requirement policy that precludes this possibility altogether. More precisely, my policy specifies that: a) if the establishment gives notice to any number of workers, it cannot rehire in the following period; and b) if the establishment does not give advance notice to any worker, it cannot fire the following period, but it can hire at will. Under such a policy, the problem of the establishment is much more complicated because it becomes dynamic. At any given time, the establishment has to decide to how many (if any) of its workers it will give advance notice. To do this, the establishment must forecast the value that the productivity shock will take in the following period. The relevant state variables for making its employment decision are its current labor force n and its current productivity shock s. If the establishment gives advance notice to some of its workers, which forces its next period employment 22 to a value n¢ below its current period employment, the establishment will not be able to rehire any workers in the following period under any of the realizations of the productivity shock s¢. It is forced to employ n¢ workers across all realizations of the productivity shock. The other alternative is not to give advance notice to any of its workers. In this case the establishment cannot contract employment below n in the following period, but it is free to hire. Thus, next period employment can be made contingent on the realization of the productivity shock s¢, as long as it is larger than the current employment level n. A profit-maximizing establishment will choose the best of both alternatives. Box 1 describes the establishments problem more formally. I am interested in describing the aggregate behavior of a large number of establishments similar to the one described so far. For this purpose, I assume that there are a large number of establishments facing the same random process for the individual productivity shock, but that the realizations of the shock are independent across establishments. I need to incorporate entry and exit of establishments, at least exogenously, because a substantial probability of exit will affect how establishments change their employment levels in response to their productivity shocks. For this purpose, I assume that the transition matrix Q for the individual productivity shocks is such that: 1) starting from any initial value, st reaches zero in finite time with probability one; and 2) once st reaches zero, there is zero probability that st will receive a positive value in the future. Given these assumptions, a zero value for the productivity shock can be identified with the death of an establishment. While establishments exit, v new establishments are exogenously created every period. The distribution of new establishments across productivity shocks is given by y. The new establishments can hire workers freely during their first period of activity since they are created with zero previous period employment. The state of the economy is described by a distribution xt, defined over idiosyncratic productivity shocks s and employment levels n. Equation 4 in box 2 describes the law of motion for xt in formal terms. Intuitively, the next periods number of establishments with a productivity shock s¢ and employment level n¢ is given by the sum of two terms. A first term gives the number of establishments that transit from their current shock s to the shock s¢ and choose an employment level n¢. A second term includes all new establishments born with a shock s¢, in case the unrestricted employment level that corresponds to the shock s¢ is given by n¢. In this article, I concentrate on steady state equilibria, where the distribution xt is invariant over time. 1Q/2002, Economic Perspectives BOX 1 Establishments problem Assume that the maximum present value of profits that can be attained starting from the state (n,s) is given by V(n,s). If the establishment gives advance notice to some of its workers, the best it can do is given by the following problem: 2) F (n, s ) max{sn H wn na n 4 V (na, s a )Q( s, s a )}. 1 1 i s a If the establishment does not give advance notice to any of its workers, the best it can do is given by: 3) H (n, s ) max {sn H wn na( sa) r n 4 V (na ( sa ), s a )Q( s, s a )}. 1 1 i s a If V(n,s) indeed describes the present value of profits under the optimal employment plan, it must be equal to the maximum of the two alternatives given by equations 2 and 3. Thus, the value function V satisfies the following functional equation: V (n, s ) max{F (n, s), H (n, s)}. In computations, I restrict s to take a finite number of possible values. The value function V is obtained by iterating on this functional equation starting from some initial guess. Given the invariant distribution x across establishment types, aggregate production c and aggregate employment h are given by summing production and employment across all establishments described by the distribution x. Formally, aggregate production and employment are given by equations 5 and 6 in box 2, respectively. My purpose here is to obtain quantitative estimates of the effects of advance notice requirements. For these estimates to be meaningful, the parameters of the model must reproduce important empirical observations (a procedure known as calibration). Although the article is concerned with European labor market institutions, I choose to replicate observations for the U.S. economy because this a common benchmark in applied studies. Since there are neither advance notice requirements nor firing taxes in the U.S. economy, I use a laissez-faire version of the model to reproduce U.S. observations. These observations are from the National Income and Product Accounts and establishments dynamics data Federal Reserve Bank of Chicago reported by Davis and Haltiwanger (1990). The calibration procedure, which is similar to the one provided in Veracierto (2001), is described in box 3.4 Next, I describe the effects of introducing advance notice requirements of three months duration (the length of the model period) to the partial equilibrium model calibrated above. Since I have chosen parameters to reproduce U.S. observations, the experiments provide estimates of the effects of introducing advance notice requirements in the U.S. economy. Table 2 (on page 25) reports the results. The first column reports statistics for the economy without interventions (which have been normalized at 100), and the second column reports statistics for the economy with advance notice requirements. The variables are aggregate production c, wages w, and aggregate employment h. There are two opposite effects of advance notice requirements on employment. On one hand, when establishments receive bad shocks, they cannot instantly contract their employment levels because they must give advance notice first. This tends to increase employment. On the other hand, when establishments receive positive shocks, they are less willing to hire workers, because if the shock is reversed, during the advance notice period they will be stuck with workers they dont need. This tends to lower employment. Table 2 shows (last row, second column) that this last BOX 2 Aggregation The law of motion for the distribution xt is described by: 4) xt 1 (na, s a ) 4 {( n , s ): g ( n , s , s a ) n a } Q( s, s a ) xt (n, s ) vZ ( s a )D(na, s a ), where g(n,s,s¢) is the next period employment level chosen by an establishment with current employment n and shock s when the realized next period productivity is s¢; and where c(n¢, s¢) is an indicator function that is equal to one if g(0, s¢, s¢) = n¢ (and zero otherwise). Given the invariant distribution x that satisfies equation 4 for all time periods t, steady state aggregate production c is given by: 5) c 4 sn H x(n, s ) n, s and steady state aggregate employment h in turn is given by: 6) I 4 nx( n, s ). ( n, s ) 23 BOX 3 Calibration I choose the interest rate to reproduce an annual rate of 4 percent, which is a compromise between the return on equity and the return on short-term debt (see Mehra and Prescott, 1985). This is also the value commonly used in the real business cycle literature. Since the model period is one quarter, i is selected to be 0.01. When no government regulations are imposed, equation 1 shows that the curvature parameter g in the production function determines the share of output that is paid to labor. As a consequence, it is selected to be 0.64, which is the share of labor in the national income accounts. I choose the wage rate w, in turn, to reproduce an average establishment size equal to 60 workers, which is consistent with Census of Manufacturers data. On the other hand, I select the number of establishments created every period n to generate a total employment level equal to 80 percent of the population, roughly the fraction of the working age population that is employed in the U.S. economy. I restrict the stochastic process for the productivity shocks to be a finite approximation to the following process. Realizations of the shock take values in the set: Q(0,{0}) 1 Q( s,[1, s% ]) 1 N Pr{(a S ln s Fa ) <1, s% >}, for s, s% r1, where a, r, and m are constants and e¢ is an i.i.d. (independently and identically distributed) normally distributed shock with mean zero and standard deviation s. With this functional form for the transition function, there are four parameters to be determined: m, a, r, and s. In addition, I must choose the distribution y across idiosyncratic shocks. Since all these parameters are important determinants of establishment dynamics in the model, they are selected to reproduce observations about establishment dynamics. The observations used to calibrate these parameters are the employment size distribution reported by the Census of Manufacturers, the job creation and destruction rates reported by Davis and Haltiwanger (1990), and the five-year exit rate of manufacturing establishments reported by Dunne, Roberts, and Samuelson (1989).1 The size distribution and the job creation and destruction statistics for the U.S. economy are displayed in table B1. The parameter values used to match these observations are reported in the appendix. Since the computations require a finite number of shocks and only nine employment ranges are reported in Census of Manufacturers data, nine values for the idiosyncratic shocks are used in the article. 1 8 {0} <1, d) and the transition function Q is assumed to be of the following form: Table B1 Statistics for U.S. and model economy A. U.S. economy Average size = 60% Job creation due to births = 0.62% Job creation due to continuing establishments = 4.77% Employment 5–9 10–19 20–49 50–99 100–249 Shares (%) 23.15 22.82 24.83 12.59 10.05 B. Model economy Average size = 59.6% Job creation due to births = 0.72% Job creation due to continuing establishments = 4.80% Employment 5-9 10–19 20–49 50–99 100–249 Shares (%) 26.19 31.67 20.21 13.01 3.92 Exit rate = 36.2% Job destruction due to deaths = 0.83% Job destruction due to continuing establishments = 4.89% Employment 250–499 500–999 1,000–2,499 >2,500 Shares (%) 3.86 1.68 0.73 0.28 Exit rate = 38.5% Job destruction due to deaths = 0.72% Job destruction due to continuing establishments = 4.80% Employment 250–499 500–999 1,000–2,499 >2,500 Shares (%) 2.25 2.13 0.59 0.02 Source: Lazear (1990). 24 1Q/2002, Economic Perspectives TABLE 2 Partial equilibrium analysis Production Wages Employment Laissez faire Advance notice Advance notice (shirk) Firing taxes 100.00 100.00 100.00 97.71 100.00 97.75 92.59 100.00 92.97 92.62 100.00 89.41 effect is the strongest: Introducing advance notice requirements reduces employment by 2.25 percent. We also see that even though employment decreases, this is not accompanied by an increase in labor productivity. In fact, we see that output decreases by roughly the same factor as employment. The reason is that with the introduction of the advance notice requirements, establishments that receive bad shocks do not contract employment (during the first period of the shock), and establishments that receive positive shocks do not expand employment enough. Thus, labor is allocated less efficiently across establishments. General equilibrium In the partial equilibrium analysis of the previous section, advance notice requirements reduce aggregate employment quite significantly (the effects have the same sign as in Lazear, 1990, and the opposite sign to the results in Addison and Grosso, 1995). However, there is an important reason to suspect that those partial equilibrium results are not reliable: The effects are so large that prices should have been significantly affected. Therefore, instead of assuming a fixed wage rate and interest rate, invariable to policy changes, in this section I investigate the effects of advance notice requirements allowing prices to adjust to clear all markets. That is, I provide a general equilibrium analysis of advance notice requirements. My results here show that equilibrium price changes are crucial for understanding the effects of this type of policy. To formulate a general equilibrium analysis, I introduce a few modifications to the environment. The same continuum of establishments analyzed in the previous section is still responsible for production of the consumption good, but now I explicitly introduce a household sector. In particular, the economy is now populated by a continuum of ex ante identical agents, of size normalized to one. The preferences of the representative agent are given by: d E 4 C [ln c t 0 t t Federal Reserve Bank of Chicago BIt ], where ct is consumption, ht is the fraction of the population that works, a is a positive parameter governing the marginal utility of leisure, and b is a discount parameter with 0 < b < 1. I restrict the analysis to a steady state equilibrium, where the wage rate w and the interest rate i are constant over time. There are two important decisions that a household has to makehow much to consume today relative to tomorrow and how much time to spend working. Consider the consumption decision first. If the household sacrifices one unit of consumption at date t in order to buy a bond, it loses the marginal utility of consumption at date t. In return, the household obtains 1 + i units of the consumption good at date t + 1, each of which is valued according to the marginal utility of consumption at date t + 1 and discounted according to b (in terms of utility at date t). If the household makes an optimal choice, the marginal loss of this decision at date t must be equal to the marginal gain at date t + 1. Since consumption (and therefore, the marginal utility of consumption) is constant in a steady state equilibrium, it follows that the steady state interest rate 1 + i must be equal to the inverse of the discount factor b. Observe that this interest rate is not affected by the introduction of advance notice requirements. As long as the economy is in a steady state, with constant consumption, the gross interest rate must be given by 1/b. Consider now the decision of how much time to spend working versus how much to consume. If the household spends one additional unit of time working, it loses the marginal utility of leisure. In return it obtains wage payments that allow it to buy w units of the consumption good, each of which is valued according to the marginal utility of consumption. If the household maximizes utility, the marginal loss from this intratemporal decision must be equal to the marginal gain. Thus, the wage rate w must be equal to the marginal rate of substitution between consumption and leisure: 7) ac = w. Observe that in equilibrium, consumption c is given by the aggregate production of establishments (equation 5 in box 2, page 23). Also, the fraction of the population that works, h, must be equal to the demand for labor by establishments (equation 6 in box 2). In order to perform the policy experiment, I choose parameter values identical to those in the partial equilibrium section, except for a and b, which are new. These two parameters are selected to generate the same wage rate w and interest rate i as in the partial equilibrium section. The required values are a = 0.80 and b = 0.99. 25 TABLE 3 General equilibrium analysis Production Wages Employment Welfare (%) Laissez faire Advance notice Advance notice (shirk) Firing taxes 100.00 100.00 100.00 0.00 99.18 99.18 100.05 0.86 97.24 97.24 100.37 3.08 97.25 97.25 96.51 0.56 When I introduce advance notice requirements, the wage rate must change in order to restore the equality with the marginal rate of substitution of consumption for leisure. Recall that when the advance notice requirements are introduced, table 2 shows that production drops quite substantially at the initial wage rate. Since the amount of production undertaken by establishments increases monotonically with decreases in the wage rate (because they increase their demand for labor), for the equality in equation 7 to be restored, the wage rate must decrease. As a consequence, both consumption and employment fall by a smaller amount than in the partial equilibrium analysis. The first two columns of table 3 show that the general equilibrium results in fact lead to a much smaller drop in aggregate consumptiononly 0.82 percent compared with the 2.29 percent drop in the partial equilibrium framework. Given the linear relation in equation 7, we know that the wage rate must also decrease in the same proportion. What is interesting to observe in table 3 is that the fall in the wage rate is enough to leave the employment level roughly unchanged (it increases only by 0.05 percent) instead of generating the substantial decrease (of 2.25 percent) obtained in the partial equilibrium framework. Thus, the general equilibrium results lead to employment effects that are more consistent with Addison and Grosso (1995) than with Lazear (1990). Since this is a neoclassical economy, the equilibrium without interventions is Pareto optimal,5 and introducing advance notice requirements can only reduce welfare levels. In fact, advance notice requirements produce significant deadweight losses. Table 3 shows that agents in the steady state with advance notice require a 0.86 percent permanent increase in consumption in order to be indifferent with being at the laissez faire equilibrium. Advance notice and shirking behavior Although I do not model it explicitly here, it is reasonable to expect that once workers are notified that they will be fired in the following period, their 26 performance on the job will decrease considerably. To capture this effect, I assume that the productivity of workers that are given advance notice is reduced to a fraction f of that of workers that are not given advance notice. However, workers that are given advance notice are paid the same wage rate as those that are not given advance notice. (Box 4 explains the modified establishments problem and feasibility condition in detail). Given that there are no data available for the shirking parameter f, I go to the extreme and assume that it is equal to zero. In other words, I assume that workers productivity drops to zero when they are given advance notice. The third column of table 2 reports the results for the partial equilibrium framework. We see that the effects of advance notice requirements are much larger when shirking behavior is present than when it is not. The reason is clear. Since establishments that contract employment must pay wages to workers without obtaining any production from them, the advance notice requirements impose much larger penalties. As a consequence, they have a much larger effect on the demand for labor, which drops by 7.03 percent instead of 2.25 percent. The drop in consumption is also much larger, 7.41 percent instead of 2.29 percent. This is due not only to the larger drop in the labor input, but also to the fact that production is severely affected when workers are given advance notice. When we incorporate the general equilibrium effects, we see (in the third column of table 3) that the wage rate drops by such an amount that employment actually increases by 0.37 percent when the advance notice requirements are introduced. Given this increase in employment, the drop in consumption is reduced to 2.76 percent (compared with 7.41 percent in the partial equilibrium framework). It is worth mentioning that shirking behavior produces the same sign as the empirical relation between advance notice requirements and employment levels reported by Addison and Grosso (1995); however, the magnitude of the employment response is much smaller. Also, note that the welfare costs of notice requirements are much larger when shirking behavior is allowed for3.08 percent instead of 0.86 percent, representing an extremely large welfare cost.6 Advance notice requirements versus firing taxes Hopenhayn and Rogerson (1993) and Veracierto (2001) analyzed the effects of firing taxes in a framework similar to this and found large negative effects on employment, consumption, and welfare. The parameterization in this article is similar to one of the cases 1Q/2002, Economic Perspectives BOX 4 Shirking behavior In order to allow for shirking behavior, I modify the value of giving advance notice in equation 2 as follows: F ( n, s ) max{s[na G( n na )]H na n wn 4 V (na, s a )Q( s, s a )}. 1 1 i s a The only other condition that must be modified in the general equilibrium analysis of the previous section is the one for aggregate consumption, which is now given by: c 4 ( n , s ):noticeis not given 4 ( n , s ):noticeis given sn H x(n, s ) s[na G(n na )]H x(n, s ). analyzed in Veracierto (2001).7 But, for that case, Veracierto (2001) only reported the effects of firing taxes equal to one year of wages. To facilitate comparisons with the advance notice requirements analyzed in this article, I report the effects of firing taxes equal to one quarter of wages (same length as the advance notice requirements). The firing tax I consider is a tax on employment reduction, which is rebated to households as lump-sum transfers. Box 5 describes the establishments problem in detail. Essentially, the establishment has to pay a tax in the next period equal to one period of wages per unit reduction in employment, whenever next period employment n¢(s¢) is lower than the current period employment n. The partial equilibrium effects of firing taxes are reported in the last column of table 2. We see that the effects on consumption are as large as under advance notice requirements when shirking behavior is allowed for (7.38 percent versus 7.41 percent), but the effects on employment are considerably larger (10.59 percent versus 7.03 percent). The consumption results are not surprising. If shirking behavior leads to zero productivity, under advance notice requirements firms end up facing similar firing restrictions as under firing taxes. In both cases, workers who are fired make no contribution to production, while the establishment must pay their wages anyway. Certainly there is a difference between the policies: Under advance notice requirements the firing decisions must be taken in advance, while under firing taxes they can be made after the shocks are realized. However, with the high persistence of Federal Reserve Bank of Chicago the productivity shocks, this difference is unimportant and, thus, the drop in output is almost the same in both scenarios. What is important is the difference in terms of employment outcomes. Under firing taxes, when establishments receive a bad productivity shock, workers are fired right away. Under advance notice, these same workers must be employed an additional period before they can be fired. Consequently, employment is larger under advance notice requirements (when shirking behavior is allowed) than when firing taxes are introduced. When the general equilibrium effects are considered, the drop in wages is virtually the same under firing taxes and advance notice requirements (with shirking behavior). But this decrease in wages is not large enough for firing taxes to increase employment. We see that employment falls by 3.49 percent. On the contrary, employment increases by 0.37 percent under the advance notice requirements (with shirking behavior). Observe that the welfare costs of firing taxes are much smaller than those of advance notice requirements (with shirking behavior): 0.56 percent instead of 3.08 percent. The reason is that consumption drops by the same amount in both cases, but employment decreases more under firing taxes, allowing for a larger amount of leisure. We see, in table 3, that the welfare costs of firing taxes are even smaller than those of advance notice requirements when shirking behavior is not allowed for: 0.56 percent versus 0.86 percent. The reason is that when establishments receive a zero productivity shock, workers are hired an additional period to comply with the advance notice requirement. This leads to a higher employment level and a lower amount of leisure. It is interesting to note that if the advance notice requirements were waived from establishments BOX 5 Firing taxes Under firing taxes the Bellman equation of establishments becomes: V (n, s ) max{sn H wn na ( sa ) 4{V (na( sa), sa) 1 1 i s a w max[0, n na ( sa )]}Q( s, s a )}, where V(n,s) is the present value, excluding current firing taxes, of an establishment with current employment n and current shock s. This equation, together with equations 4, 5, 6, and 7, defines an equilibrium with firing taxes rebated as lump sum taxes. 27 that exit the market, the amount of employment h in the advance notice column would be 99.40 instead of 100.05 and the welfare cost of the policy would be 0.44 percent, a lower cost than that of firing taxes. When advance notice requirements generate shirking behavior, their effects can be considerably larger. However, when the general equilibrium effects are taken into account, advance notice requirements actually have a positive effect on employment. This effect is of the same sign as in Addison and Grosso (1995), but the magnitude is much smaller. In terms of welfare effects, I find that advance notice requirements are quite costlyin fact even costlier than firing taxes. While firing taxes equal to three months of wages reduce welfare by 0.56 percent, advance notice requirements lead to welfare costs that range between 0.86 percent and 3.08 percent, depending on the amount of shirking behavior generated. However, the large welfare cost of advance notice requirements allowing for shirking behavior was calculated under the assumption that workers who shirk do not obtain leisure. This is probably an unrealistic assumption and, as such, we should interpret these results with caution. While the results in this article suggest that advance notice requirements can be extremely costly, in order to provide a more definite answer they should be analyzed in a model that explicitly considers the shirking decisions. A model based on efficiency wages may provide a suitable framework of analysis. Conclusion This article analyzes the effects of advance notice requirements in a general equilibrium model of establishment-level dynamics of the type introduced by Hopenhayn and Rogerson (1993). I find that when advance notice requirements do not lead to shirking behavior, the effects of advance notice requirements are relatively small. Establishments do not tend to alter their employment levels considerably for the following reasons: a) next periods productivity is likely to be similar to current productivity (given the high persistence of the shocks); b) employment can be freely increased if a good shock occurs next period; c) employment can be decreased after one period if a bad shock occurs; and d) during the period of notice the workers remain productive. In a partial equilibrium framework, I find that advance notice requirements reduce employment, but when I consider general equilibrium effects, employment is not much affected. The reason is that the advance notice requirements lead to a substantial reduction in equilibrium wages, which sustains the employment level. APPENDIX: PARAMETERS Prices and technology i = 0.01 w = 0.3297 g s1 = 1.00 s6 = 4.19 s2 = 1.32 s7 = 5.38 s3 = 1.79 s8 = 7.30 s4 = 2.35 s9 = 10.65 y = 6.8e5 y = 0.0 y y y y = 0.64 Productivity shocks s0 = 0.00 s5 = 3.19 Distribution over initial productivity shocks y y = 9.995e1 = 0.0 5 0 y y = 2.3e4 = 0.0 6 1 2 7 = 1.6e4 = 0.0 8 3 = 0.0 = 0.0 9 4ÿ Transition matrix Q: 1.000 0.087 0.005 0.005 0.005 0.005 0.005 0.005 0.005 0.005 28 0.000 0.848 0.084 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.065 0.879 0.086 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.032 0.847 0.088 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.062 0.877 0.090 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.031 0.846 0.092 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.059 0.808 0.094 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.095 0.873 0.096 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.028 0.896 0.099 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.004 0.896 1Q/2002, Economic Perspectives NOTES 1 For an extensive discussion of dismissal regulations, see Emerson (1988) and Piore (1986). If the establishment could rehire workers that were given advance notice, then it would give advance notice to all of its workers and rehire them at will the following period, depending on the value of the establishments individual productivity. Clearly, if this were allowed, advance notice requirements would have no effect. while in this article labor is the only factor. Another difference is that in the former paper entry of establishments was endogenous, while here it is exogenous. 2 Observe that variations in the productivity of the establishment determine its employment expansion and contraction over time. 3 A Pareto optimal allocation maximizes the utility level of the representative agent within the set of feasible allocations. 5 A good part of the welfare cost of advance notice requirements when shirking behavior is allowed for is due to the assumption that workers that shirk do not enjoy leisure. 6 In particular, it corresponds to the economy without capital referred to in that paper as the H-R economy. 7 A main difference with Veracierto (2001) is that that paper had a flexible form of capital as an alternative factor of production, 4 REFERENCES Addison J., and J. Grosso, 1995, Job security provisions and employment: Revised estimates, Centre for Labour Market and Social Research, working paper, No. 95-15. Hopenhayn, H., and R. Rogerson, 1993, Job turnover and policy evaluation: A general equilibrium analysis, Journal of Political Economy, Vol. 101, No. 5, pp. 915938. Bentolila, S., and G. Bertola, 1990, Firing costs and labor demand: How bad is eurosclerosis?, Review of Economic Studies, Vol. 57, No. 3, pp. 381402. Lazear, E., 1990, Job security provisions and employment, Quarterly Journal of Economics, Vol. 105, No. 3, pp. 699726. Davis, S., and J. Haltiwanger, 1990, Gross job creation and destruction: Microeconomic evidence and macroeconomic implications, in NBER Macroeconomics Annual, Vol. 5, O. Blanchard and S. Fischer (eds.), pp. 123168. Mehra, R., and E. Prescott, 1985, The equity premium: A puzzle, Journal of Monetary Economics, Vol. 15, No. 2, pp. 145161. Dunne, T., M. Roberts, and L. Samuelson, 1989, The growth and failure of U.S. manufacturing plants, Quarterly Journal of Economics, Vol. 104, No. 4, pp. 671698. Piore, M., 1986, Perspectives on labor market flexibility, Industrial Relations, Vol. 25, No. 2, pp. 146166. Veracierto, M., 2001, Employment flows, capital mobility, and policy analysis, International Economic Review, Vol. 42, No. 3, pp. 571595. Emerson, M., 1988, Regulation or deregulation of the labour market, European Economic Review, Vol. 32, No. 4, pp. 775817. Federal Reserve Bank of Chicago 29 When can we forecast inflation? Jonas D. M. Fisher, Chin Te Liu, and Ruilin Zhou Introduction and summary The practice of forecasting inflation has generally been considered an important input in monetary policymaking. Recently, this view has come under attack. In an article that appeared in the Federal Reserve Bank of Minneapoliss Quarterly Review, Atkeson and Ohanian (2001, hereafter A&O) argue that the likelihood of accurately predicting a change in inflation using modern inflation forecasting models is no better than a coin flip. They conclude that these forecasting models cannot be considered a useful guide for monetary policy. In this article, we reexamine the findings that underlie this conclusion. We show that it may be possible to forecast inflation over some horizons and in some periods. A&O study the properties of standard Phillipscurve-based inflation forecasting models. These models relate changes in inflation to past values of the unemployment gap (the difference between unemployment and a measure of unemployment believed to be associated with non-accelerating inflation, the so-called NAIRU [non-accelerating inflation rate of unemployment]), past changes in inflation, and perhaps other variables believed to be useful indicators of inflation.1 Recently, Stock and Watson (1999, hereafter S&W) proposed a generalized version of the Phillips curve and argued that their generalization is superior to these standard models as a forecasting tool. Focusing on the one-year-ahead forecast horizon, A&O argue that unemployment-based Phillips curve models and S&W generalized Phillips curve models can do no better than a naive model, which says that inflation over the coming year is expected to be the same as inflation over the past year. This analysis focuses on the ability to forecast the magnitude of inflation in the Consumer Price Index (total CPI), the CPI less food and energy components (core CPI), and the personal consumption expenditures deflator (total PCE) over the sample period 1985 to 2000. 30 To gain some insight into these findings, figure 1, panel A displays 12-month changes in 12-month core CPI from 1967 to 2000. The vertical lines in this figure (in 1977, 1985, and 1993) divide the sample period into four periods. It is immediately clear that in the two later periods, that is, the sample period considered by A&O, the volatility of changes in inflation was much lower than in the two earlier periods. This change in the behavior of inflation seems to be coincident with the change in monetary policy regime that is generally thought to have taken effect in the mid-1980s.2 The lower volatility and the possibility of a changed monetary policy regime in the later two sample periods may favor the naive model studied by A&O. Figure 1, panel B shows that PCE less food and energy components (core PCE) behaves in a similar fashion. These changes in the behavior of inflation raise the question of whether A&Os findings are due to special features of the data in the sample period they chose to focus on. To address this possibility, we extend the A&O analysis by studying three distinct sample periods, 197784, 198592, and 19932000. In addition, we add core PCE inflation to the list of inflation measures and we consider a broader class of StockWatson type models. A&O focus on the oneyear forecast horizon. Given the lags inherent in the effects of monetary policy actions, it is reasonable to consider whether their results extend to longer horizons. Consequently, we analyze both the one-year and two-year forecast horizons. Our findings confirm the A&O results for the 19852000 period, but not for 197784. The Phillips curve models perform poorly in both the 198592 and 19932000 periods when forecasting core CPI. Jonas D. M. Fisher is a senior economist, Chin Te Liu is an associate economist, and Ruilin Zhou is a senior economist at the Federal Reserve Bank of Chicago. 1Q/2002, Economic Perspectives FIGURE 1 12-month changes in 12-month core inflation A. Core CPI 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 1967 '69 '71 '73 '75 '77 '79 '81 '83 '85 '87 '89 '91 '93 '95 '97 '99 '01 '71 '73 '75 '77 '79 '81 '83 '85 '87 '89 '91 '93 '95 '97 '99 '01 B. Core PCE 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 1967 '69 Source: Haver Analytics, Inc., 2001, U.S. economic statistics database, July. However, when forecasting core PCE, these models improve significantly relative to the naive model in the 19932000 period. While the Phillips curve models do poorly for the one-year-ahead forecast horizon, we do find evidence in favor of the Phillips curve models for the two-year-ahead forecast horizon, at least with respect to core inflation. Taken together, these findings are consistent with our suspicion that periods of low inflation volatility and periods after regime shifts favor the naive model. Federal Reserve Bank of Chicago The relatively poor performance of the Phillips curve models reflects their inability to forecast the magnitude of inflation accurately. Ultimately, the way we assess our forecasting models should reflect the usefulness of the forecasts in policymaking. In our view, policymakers understand that precise forecasts of inflation are fraught with error. As a result, they pay considerable attention to the direction of change of future inflation. For this reason, we do not view measures of forecast performance used by A&O and 31 many others that emphasize magnitude as the only criteria for evaluating forecasting models. Consequently, we consider a complementary approach to evaluating forecasting models that emphasizes the forecasted direction of change of future inflation. Under the assumption that forecast errors are symmetrically distributed about the forecast, the naive model provides no information about future inflation; it is no better than a coin flip at predicting the future direction of inflation. Under the same symmetry assumption, the Phillips curve models predict that inflation will change in the direction indicated by comparing the point forecast with the current level of inflation. We analyze the ability of our Phillips curve models to forecast the direction of inflation and find that they do quite well. Over the entire 19772000 period, the Phillips curve models are able to forecast the correct direction of inflation one year ahead between 60 percent and 70 percent of the time. For the same period, the models forecast the correct direction two years ahead more than 70 percent of the time. These results suggest that the Phillips curve models forecast the direction of inflation changes relatively well across measures of inflation and across time. But when it comes to forecasting the magnitude of inflation changes, there may be times, such as after a change in monetary policy regime, when the naive model may do better than the Phillips curve models. The last question we address is whether it is possible to improve on the forecasts of the naive model in difficult times by using the directional information contained in the Phillips curve models. We show that it is possible to improve on the naive model, although the improvement is modest. One interpretation of our findings is that it is possible to forecast inflation accurately during some periods, but not others. We argue that the periods in which it is difficult to forecast inflation are associated with changes in monetary policy regime, broadly interpreted. This implies that if we are in a stable monetary regime and expect the regime to persist, then it may make sense for policymakers to pay attention to inflation forecasts. In the next section, we outline the different forecasting models that we consider in our analysis. Next, we discuss the standard methodology we implement to evaluate the ability of these models to forecast the magnitude of future inflation. We then discuss our results for forecasting magnitude and present our analysis of forecasting directional changes in inflation. We describe our procedure for combining the naive model with our directional forecasts and how well this procedure performs over our sample period. Finally, we discuss some possible policy implications of our findings. 32 Statistical models of inflation The standard approach to forecasting inflation is rooted in ideas associated with the Phillips curve, the statistical relationship between changes in inflation and measures of overall economic activity. The generalized version of the Phillips curve proposed by S&W involves variables that summarize the information in a large number of inflation indicators. S&W argue that their generalization is superior to conventional Phillips curves as a forecasting tool. A&O argue that neither the conventional nor the generalized Phillips curve framework can do better than a simple forecast (their naive model) that says inflation over the coming year is expected to be the same as inflation over the past year. We reexamine this claim using a broader class of S&W-type models than considered by A&O. Now we describe in detail the models we study. The naive model The benchmark for evaluating our models is the naive model described by A&O. The starting point for the naive model is the martingale hypothesis, which states that the expected value of inflation over the next 12 months is equal to inflation over the previous 12 months. Specifically, 1) 12 Et Q12 t 12 Q t , where the 12-month inflation rate, Q12 t is defined as the 12-month change in the natural logarithm of the price indexes pt, Q12t ln pt ln pt 12 , and Et denotes the expectation conditional on date t information. The naive model equates the forecast of inflation over the next 12 months, Qˆ 12 t 12 , with its conditional expectation. That is, 12 2) Qˆ 12 t 12 Q t . Notice that if the martingale hypothesis holds, then the expected value of 12-month inflation in the second year following date t must also equal inflation over the 12 months prior to date t, that is 12 Et Q12 t 24 Q t . Similar to the 12-month forecast, the naive model equates the forecast of inflation over the next 24 months, Qˆ 12t 24 , with its conditional expectation: 1Q/2002, Economic Perspectives 3) Qˆ 12t 24 Q12t . Generalized Phillips curve models The simplest alternative to the naive model postulates that changes in 12-month inflation only depend on recent changes in one-month inflation. That is, for J = 12, 24, 12 4) Q12 t J Qt B C( L ) (Q t Q t 1 ) F t J , where the one-month inflation rate, pt, is defined by pt = ln pt ln pt1. In addition, et is an error term, and b(L) specifies the number of lags in the equation.3 Below, we refer to this as model 1. The next model we consider is based on the Chicago Fed National Activity Index (CFNAI). This index is a weighted average of 85 monthly indicators of real economic activity. The CFNAI provides a single, summary measure of a common factor in these national economic data. As such, historical movements in the CFNAI closely track periods of economic expansion and contraction. The index is closely related to the Activity Index studied in S&W.4 Our model based on this index postulates that changes in 12-month inflation, in addition to recent changes in inflation, also depend on current and past values of the CFNAI. That is, for J = 12, 24, 5) Q12t J Q12t B C( L) (Q t Q t 1 ) H ( L)at F t J , where at denotes the value of the CFNAI at date t, and b(L) and g(L) specify the number of lags in inflation and the index, respectively, included in the equation. We refer to this as model 2. The remaining models we consider are based on the diffusion index methodology described in S&W. This methodology uses a small number of unobserved indexes that explain the movements in a large number of macroeconomic time series. Our implementation of the S&W methodology uses 154 data series, including data measuring production, labor market status, the strength of the household sector, inventories, sales, orders, financial market, money supply, and price data. The procedure that obtains the indexes processes the information in the 154 series so that each index is a weighted average of the series and each index is statistically independent of the others. We consider six indexes, d1t, d2t, ..., d6t, which are ranked in descending order in terms of the amount of information embedded in them. Federal Reserve Bank of Chicago Our diffusion index models postulate that changes in 12-month inflation depend on recent changes in inflation, and current and past values of a number of diffusion indexes. That is, for J = 12, 24, 12 6) Q12 C( L) (Q t Q t 1 ) t J Qt B ¤ R ( L)d K i 1 i it Ft J , where K = 1, 2, ..., 6, and b(L) and qi(L) specify the number of lags in inflation and diffusion index i, respectively, included in the equation. As more indexes are included in the equation, more information about the 154 series is incorporated in the forecast. We refer to these as models 3, 4, ..., 8. For all these models, we equate the forecasts of inflation with the conditional expectation implied by the model. That is, for J = 12, 24, Qˆ 12t J Et Q12t J . We estimate all these models by ordinary least squares (OLS). In each case, we use the Bayes Information Criterion (BIC) to select the number of lags of inflation, the CFNAI, and the diffusion indexes. Intuitively, BIC selects the number of lags to improve the fit of the model without increasing by too much the sampling error in the lag coefficients. We allowed for the possibility that lags could vary from 0 to 11. In real time, it is difficult to choose the appropriate model to use to form a forecast. To address this issue, we consider a forecasting model in which the forecast of inflation at any given date is the median of the forecasts of models 1 through 8 at that date.5 This procedure has the advantage that it can be applied in real time. We call this the median model. Stock and Watson (2001) use a similar model. For convenience, we refer to the collection of models comprising models 1 through 8 plus the median model as Phillips curve models. Model evaluation methodology We evaluate the accuracy of the generalized Phillips curve models by comparing them with the naive model. We do this through various simulated out-of-sample forecasting exercises. These exercises involve constructing inflation forecasts that a model would have produced had it been used historically to generate forecasts of inflation. Two drawbacks of our approach, which also affect A&O and S&W, are that we do not use real-time data in our forecasts and we assume all the data are available up to the forecasting date. On a given date, particular data series may not 33 yet be published. Also, many data series are revised after the initial release date. In our forecasting exercises, we calculate the CFNAI and diffusion indexes assuming all the series underlying the indexes are available up to the forecast date.6 In practice, this is never the case and we must fill in missing data with estimates. Since we do not use real-time data to construct the CFNAI and diffusion indexes, we also abstract from problems associated with data revisions. We suspect that these drawbacks lead us to overstate the effectiveness of our CFNAI and diffusion index models. Data revision is also a problem for the lagged inflation and naive PCE models, since this price index is subject to revisions. It does not affect the CPI versions of these models, since the CPI is never revised. To assess the accuracy of our various models, we first construct a measure of the average magnitude of the forecasting error. The measure we use is root mean squared error (RMSE). The RMSE for any forecast is the square root of the arithmetic average of the squared differences between the actual inflation rate and the predicted inflation rate over the period for which simulated forecasts are constructed. For J = 12, 24, 7) ¥1 ¤ ª¨Q §T RMSE ¦ T t 1 12 t J ´ Qˆ 12t J ¹· µ ¶ 2 1/ 2 , where T denotes the number of forecasts made over the period under consideration. We compare the forecast of a given Phillips curve model with that of the naive model by forming the ratio of the RMSE for the Phillips curve model to the RMSE for the naive model. We call this ratio the relative RMSE. A ratio less than 1 thus indicates that the Phillips curve model is more accurate than the naive model. Subtracting 1 from the ratio and multiplying the result by 100 gives the percentage difference in RMSE between the two models. The RMSE might be strongly affected by one or two large outliers. We reworked our analysis using a measure of forecasting error that places equal weight on all forecasting errors and found that our findings are robust.7 The RMSE statistics are subject to sampling variability and, consequently, are measured with error. In principle, we could use Monte Carlo methods to assess the magnitude of this error. However, this would require specifying an underlying datagenerating process for all the variables in our analysis (more than 150 of them). One should keep this sampling error in mind when interpreting the results below. The sample period of our analysis begins in 1967. We chose this date because it is the beginning date for the data used to construct the CFNAI and the diffusion indexes.8 We estimate the forecasting equations using 34 rolling regressions, a method that keeps the number of observations in the regression constant across forecasts. Since it excludes observations from the distant past, this approach can in principle accommodate the possibility of structural change in the data-generating process. We choose this sample length for the rolling regression procedure to be 15 years.9 Finally, we consider three distinct periods over which to evaluate the forecasts of the models: 1977 84, 198592, and 19932000. To compare our results with those in A&O, we also evaluate the overall performance of the models over the 19852000 period. To complete the analysis, we study the performance of the models over the entire 19772000 period as well. The 197784 period is one of high inflation volatility and general economic turbulence. The 198592 period is generally associated with a new monetary policy regime. This period also includes a mild recession. The 19932000 period witnessed uninterrupted economic expansion, stable monetary policy, and declining inflation. Atkeson and Ohanian revisited We estimated the Phillips curve models for the five sample periods and computed the associated RMSEs and the relative RMSEs. For models 18, we do not report all the results, just the results for the best models. We do this to demonstrate the potential forecasting capacity of these models. A&O report the performance of the best and worst models they look at across different lag lengths. S&W use BIC to select lag length and report the performance of all their models. All of these approaches suffer from the deficiency that in real time one may not know which is the best performing model. Our median model overcomes this deficiency. Table 1 displays the RMSE statistics of the best and median 12-month-ahead and 24-monthahead forecasts for the five sample periods and four measures of inflation. The table also identifies the best performing models. The numbers in bold in the table indicate cases in which the naive model outperforms the Phillips curve models. Finally, for each case we report the RMSE for the naive model. Regarding the 12-month-ahead forecasts in table 1, our findings are as follows. First, over the 19852000 period, essentially all the relative RMSEs are at least as large as 1. That is, the naive model outperforms all the Phillips curve models. This finding confirms the result reported in A&O that Phillips curve models have not performed well over the last 15 years. Second, while inflation forecasting appears to have been quite difficult over the last 15 years, for core PCE it has become a little easier in the most recent forecasting 1Q/2002, Economic Perspectives TABLE 1 Forecasting the magnitude of inflation: Phillips curve models vs. naive model 12 months ahead 24 months ahead Sample period Naive model RMSE Core CPI 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 2.360 0.667 0.341 2 1 2 0.768 0.985 1.110 0.885 1.290 1.181 3.802 0.780 0.705 4 1 2 0.930 0.894 0.765 0.868 1.615 0.768 1985:01–2000:12 1977:01–2000:12 0.530 1.430 1 2 1.016 0.891 1.268 0.927 0.744 2.278 1 5 0.903 1.000 1.304 0.906 Core PCE 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 1.238 0.481 0.514 2 1 4 0.954 1.409 0.750 1.033 1.412 0.749 2.100 0.617 0.802 5 1 4 0.887 1.221 0.532 0.765 1.197 0.542 1985:01–2000:12 1977:01–2000:12 0.498 0.822 1 2 1.188 1.048 1.109 1.052 0.716 1.346 6 5 0.933 0.902 0.847 0.781 Total CPI 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 2.674 1.489 0.716 2 1 1 0.687 0.982 1.085 0.765 0.982 1.193 4.525 1.695 1.032 4 1 1 0.744 0.981 1.035 0.696 1.245 1.002 1985:01–2000:12 1977:01–2000:12 1.168 1.815 1 2 1.002 0.865 1.025 0.845 1.403 2.853 1 6 0.996 0.954 1.184 0.795 Total PCE 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 1.705 1.025 0.633 2 1 4 0.841 0.978 0.953 0.953 1.012 0.960 2.977 1.102 0.924 6 1 6 0.751 1.029 0.773 0.686 1.279 0.772 1985:01–2000:12 1977:01–2000:12 0.852 1.205 1 2 1.003 0.974 0.998 0.968 1.017 1.909 1 6 1.020 0.909 1.098 0.781 Best performing model Rel. RMSE Median rel. RMSE Naive model RMSE Best performing model Rel. RMSE Median rel. RMSE Notes: Fifteen-year rolling regression. RMSE is root mean squared error. Numbers in bold indicate cases in which the naive model outperforms the Phillips curve models. period. In particular, the forecast by the best model and the median forecast have RMSEs 25 percent lower than the naive model over the 19932000 period. Note, however, that this pattern is not true for core CPI. Third, the Phillips curve models are generally better than the naive model in the 197784 period. This result is uniform across inflation measures, except for the median forecast for core PCE. In some cases the improvement is quite dramatic. For example, the best CFNAI model is more than 30 percent better than the naive model when forecasting total CPI. Even the median forecast is about 24 percent better than the naive model. The results for the 24-month-ahead forecasts in table 1 suggest that, over longer horizons than 12 months, the Phillips curve models may more consistently outperform the naive model. In particular, the best models at forecasting core inflation outperform the naive model in the 19852000 period. However, the gains are not dramatic. For core CPI, the gain is roughly 10 percent, and for core PCE it is about 7 Federal Reserve Bank of Chicago percent. The median forecasts for core CPI over this period fare worse, but they are better for core PCE with a gain of 15 percent over the naive model. In the recent 19932000 period, the gains over the naive model are more substantial. The best models improve relative to the naive model by 24 percent for core CPI and 47 percent for core PCE. We see similar gains for the median forecasts. Finally, there are across the board gains using Phillips curve models to forecast 24 months ahead for the 197784 period. We can summarize these findings as follows. First, the naive model does poorly in the 197784 period and relatively well in the 198592 period, forecasting 12 months ahead. Second, the naive model does not do well forecasting PCE inflation in the recent 1993 2000 period. Finally, the naive model does better forecasting 12 months ahead than 24 months ahead. We can attribute the first finding to an apparent structural change in the early 1980s and the consequent decline in inflation volatility in the post-1984 period compared with the previous period. This decline in 35 volatility is evident in the pattern of RMSEs for the naive model in table 1 (also see figure 1).10 Given that the naive model predicts no change in inflation, it should do better in a period of low inflation volatility than in a period of high volatility. It is unclear how the performance of the Phillips curve models is affected by inflation volatility. Nevertheless, we suspect that changes in inflation volatility are a contributing factor to the poor performance of the naive model in the 197784 period and its significant improvement in the most recent 15 years. Another factor that probably plays an important part in explaining our first finding is that forecasting models do relatively well in a stable environment. If the structure of the economy changes, then regression equations tend to forecast with more error. We suspect the change in structure in the early 1980s has a lot to do with a change in monetary policy regime around that time. We think volatility and structural stability change may explain the second finding as well. In particular, it appears that there was a further decline in core CPI volatility in the 1993 2000 period, which is not matched by core PCE. We think one possible explanation of the improved performance of the Phillips curve models at the 24-month forecast horizon has to do with the sluggish response of the economy to monetary policy actions. It is generally understood that economic activity and inflation respond with a considerable lag to changes in monetary policy, and that inflation is more sluggish in its response than economic activity. If this is true then there may be less information about future inflation in the 12-month-ahead forecasts than in the 24-month-ahead forecasts. Note that as the forecast horizon is increased, forecasting performance in terms of RMSE generically worsens. We can see this by comparing the RMSEs of 12-month-ahead and 24-month-ahead forecasts of the naive model in table 1. Evidently, the forecast errors for the Phillips curve models deteriorate at a slower rate than the forecast errors for the naive model. Forecasting direction In the previous section we used the RMSE criterion to evaluate the models. This measure emphasizes the ability of a forecasting model to predict the magnitude of inflation. In this section, we consider a complementary approach to evaluating forecasting models, which emphasizes the forecasted direction of change of future inflation. What do the models we have described have to say about direction of change of inflation? First, consider the naive model. Strictly speaking, according to equations 2 and 3, this model always predicts no change 36 in inflation. In principle the martingale hypothesis, equation 1, on which the naive model is based, could be used to make forecasts about direction. Given the conditional distribution of inflation 12 months and 24 months ahead, we could assess the probability of an increase or decrease in inflation over these horizons and use this to make predictions about the direction of change. If this distribution is symmetric around the conditional mean, then the martingale hypothesis would suggest that the likelihood of an increase in inflation is always 50 percent. If the distribution is skewed, the odds of inflation changing in a particular direction would be better than a coin flip. The martingale hypothesis does not provide any information about the nature of the conditional distribution. Deriving predictions about the direction of inflation changes from a Phillips curve model is more straightforward. The main difference from the naive model is that the conditional expectation of inflation 12 months and 24 months ahead is not constrained to equal current inflation. Consequently, we can infer the direction of change by making minimal assumptions about the distribution of the error terms in equations 46. Specifically, if these distributions are symmetric, then the direction of change is given by the sign of the difference between the conditional forecast and the current value of inflation. Now we analyze the ability of our models to forecast direction. We assume the forecast errors are symmetrically distributed. Therefore, the naive model predicts inflation increases with probability 50 percent. We evaluate our Phillips curve models by assessing how well they can forecast direction relative to this baseline. Specifically, for a given Phillips curve model, let Dˆ t12J be the predicted direction of change in inflation J periods ahead. We define Dˆ t12J as follows for J = 12, 24: 8) « 1 if Et Q12t J Q12t Dˆ t12J ¬ , 1 otherwise where Dˆ t12J 1 indicates a forecasted increase in inflation and Dˆ t12J 1 indicates a decrease. Actual changes in inflation are defined analogously. Let Dt12 J be the actual direction of change in inflation J periods ahead, for J = 12, 24, « 1 if Q12t J Q12t . 1 otherwise Dt12J ¬ We measure the directional change performance of a model by measuring the percentage of the directional 1Q/2002, Economic Perspectives TABLE 2 Forecasting the direction of inflation changes 12 months ahead Best performing model Sample period 24 months ahead PDPC Median PDPC Best performing model PDPC Median PDPC Core CPI 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 3 7 4 75.0 62.5 78.1 71.9 59.4 80.2 3 6 1 91.7 66.7 85.4 82.3 63.5 78.1 1985:01–2000:12 1977:01–2000:12 7 7 70.3 69.1 69.8 70.5 1 3 74.5 75.0 70.8 74.7 Core PCE 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 2 1 8 79.2 61.5 69.8 69.8 42.7 69.8 2 6 1 90.6 61.5 90.6 87.5 52.1 82.3 1985:01–2000:12 1977:01–2000:12 1 2 64.1 67.0 56.3 60.8 6 2 70.8 72.2 67.2 74.0 Total CPI 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 2 8 4 86.5 60.4 60.4 71.9 58.3 57.3 4 6 3 92.7 76.0 77.1 89.6 72.9 74.0 1985:01–2000:12 1977:01–2000:12 5 2 59.4 62.8 57.8 62.5 5 4 73.4 78.5 73.4 78.8 Total PCE 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 3 1 5 89.6 56.3 71.9 77.1 52.1 67.7 4 4 5 94.8 68.8 80.2 93.8 62.5 76.0 1985:01–2000:12 1977:01–2000:12 7 7 62.0 65.6 59.9 65.6 5 5 74.0 80.2 69.3 77.4 Notes: Fifteen-year rolling regression. RMSE is root mean squared error. PDPC indicates percentage of directional predictions that are correct. Numbers in bold indicate failure with respect to the naive model. predictions that are correct (PDPC) in a particular sample period. This percentage is defined as (for J = 12, 24), PDPC 1 T ¤ I \Dˆ T t 1 12 t J Dt12J ^ , where I takes the value 1 when its argument is true (that is, Dˆ t12J Dt12J ), and 0 otherwise. We used our estimates of the Phillips curve models computed for the RMSE comparisons in the previous section to make predictions about the direction of change of inflation according to equation 8. We report the findings for the best Phillips curve models and for the median model. Table 2 displays the 12- and 24-month-ahead directional predictions for the five sample periods and four measures of inflation. The table also identifies the best performing models. Numbers in bold indicate failure with respect to the naive model. Federal Reserve Bank of Chicago Our findings can be summarized as follows. It is immediately clear from the tables that the Phillips curve models predict direction in excess of 50 percent of the time for both 12-month and 24-month horizons in all but one case. Similar to their performance in terms of RMSE, these models are typically best at predicting directional change during the 197784 period and worst in the 198592 period. Interestingly, the best models at predicting directional changes are not the same as the best models in terms of RMSE. For example, model 4 (the model that includes d1t and d2t) provides the best 12-month-ahead forecasts of directional changes of core CPI over the 19932000 period. In terms of RMSE, model 2 provides the best forecasts over this sample period. Moreover, it is possible for a model to do well on directional changes while underperforming the naive model in terms of magnitude. In the example just given, the best directional change model is correct more than 78 percent of the time, but the best RMSE models in the corresponding period 37 FIGURE 2 Median model directional predictions of 12-month changes in 12-month core inflation A. Core CPI—70.5% correct 6.0 4.0 2.0 0.0 -2.0 Correct Incorrect -4.0 -6.0 1977 '79 '81 '83 '85 '87 '89 '91 '93 '95 '97 '99 '01 '97 '99 '01 B. Core PCE—60.8% correct 6.0 4.0 2.0 0.0 -2.0 Correct Incorrect -4.0 -6.0 1977 '79 '81 '83 '85 '87 are worse than the naive model. Finally, the 24-monthahead directional change forecasts perform better than the 12-month ahead forecasts. Figures 2 and 3 provide information on when our directional change forecasts are correct. The bars indicate actual changes in core CPI and PCE inflation and the green bars indicate the correct directional predictions of the median model over the 19772000 period. The main lesson from these figures is that much of the success of the directional forecasts derives from periods in which there is a consistent trend in one 38 '89 '91 '93 '95 direction or the otherthe longer periods of consecutive increasing or decreasing inflation are associated with better directional forecasting. The relatively poor performance in the 198592 period may be partially due to the absence of a trend. As with our interpretation of the RMSE findings, we believe the change in monetary policy regime may also play a role. Interestingly, in the recent 19932000 period, despite the general downward trend in core CPI inflation, the one-year directional forecasts are able to correctly anticipate the brief episodes of increasing inflation. 1Q/2002, Economic Perspectives FIGURE 3 Median model directional predictions of 24-month changes in 12-month core inflation A. Core CPI—74.7% correct 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 Correct Incorrect -6.0 -8.0 1977 '79 '81 '83 '85 '87 '89 '91 '93 '95 '97 '99 '01 '99 '01 B. Core PCE—74.0% correct 8.0 6.0 4.0 2.0 0.0 -2.0 -4.0 Correct Incorrect -6.0 -8.0 1977 '79 '81 '83 '85 '87 Can we improve on the naive model in difficult times? Confirming the A&O findings, we show that the naive model has done quite well over the last 15 years in forecasting the magnitude of inflation. Over the same period, the Phillips curve models seem to provide information on the direction of changes in inflation. A natural question is whether we can combine these models to get a better forecast for magnitude. Intuitively, we should be able to do this by shaving the naive model forecasts up or down Federal Reserve Bank of Chicago '89 '91 '93 '95 '97 according to the directional predictions. In this section, we explore this idea and show that, indeed, it is possible to do somewhat better than the naive model. We modify the naive model by adjusting its forecast in the direction predicted by a given Phillips curve or median model. That is, for J = 12, 24, Qˆ 12t J Q12t Dˆ t12J s vt , where Dˆ t12J is defined in equation 8 and vt is the adjustment factor. The intuition is that, for small enough 39 vt, we should be able to improve on the naive model. In addition, we believe the magnitude of vt should be related to recent changes in inflation. Consequently, we adjust the naive model by a percentage of the average inflation change in the recent past. That is, we assume vt evolves as follows: vt M s ¤ t j t N Q12j Q12j 12 . There is nothing in this approach that pins down vt, and one may define vt in other ways, provided that it is not too large. This formulation assumes symmetry in magnitude of increases and decreases in inflation. Choices of l and N reflect the forecasters belief in the relevance of past volatility of inflation for future volatility. For fixed N, intuition suggests that, for small enough l, there will be at least a slight improvement over the naive model. We choose l = 0.1 and N to correspond to the beginning of the regression sample. We call this the combination model. Table 3, constructed in the same way as table 1, shows how well the combination model performs relative to the naive model. These results confirm our belief that we can improve on the naive model almost uniformly. For example, over the 19852000 period, the improvement of the best performing combination model for core CPI is about 7 percent and that for core PCE is about 3 percent for the 12month horizon. For the 24-month horizon, the gains are 9 percent and 6 percent, respectively. Admittedly, these are not large improvements. The results for the median-based combination forecasts are less encouraging. The bad performance in the 198592 period seems to be driven by the relatively poor performance of the directional forecasts for both one-year and two-year horizons. The performance of the combination model may improve slightly by increasing l by a small amount, but not by much. TABLE 3 Forecasting the magnitude of inflation: Combination models vs. naive model 12 months ahead 24 months ahead Sample period Naive model RMSE Core CPI 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 2.360 0.667 0.341 2 7 7 0.959 0.955 0.855 0.968 0.981 0.859 3.802 0.780 0.705 3 6 1 0.950 0.965 0.833 0.963 0.994 0.854 1985:01–2000:12 1977:01–2000:12 0.530 1.430 7 2 0.935 0.965 0.957 0.967 0.744 2.278 1 3 0.910 0.950 0.934 0.961 Core PCE 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 1.238 0.481 0.514 5 1 1 0.954 0.992 0.944 0.962 1.069 0.942 2.100 0.617 0.802 3 6 1 0.945 0.970 0.910 0.950 1.016 0.925 1985:01–2000:12 1977:01–2000:12 0.498 0.822 1 7 0.967 0.966 1.003 0.972 0.716 1.346 6 3 0.943 0.948 0.960 0.952 Total CPI 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 2.674 1.489 0.716 3 8 5 0.951 0.983 0.990 0.959 0.990 0.997 4.525 1.695 1.032 4 6 5 0.949 0.952 0.915 0.956 0.969 0.938 1985:01–2000:12 1977:01–2000:12 1.168 1.815 6 2 0.987 0.965 0.992 0.968 1.403 2.853 6 6 0.950 0.952 0.961 0.957 Total PCE 1977:01–1984:12 1985:01–1992:12 1993:01–2000:12 1.705 1.025 0.633 3 1 5 0.939 0.988 0.944 0.958 1.003 0.962 2.977 1.102 0.924 5 5 5 0.941 0.934 0.911 0.946 0.961 0.925 1985:01–2000:12 1977:01–2000:12 0.852 1.205 7 2 0.982 0.961 0.992 0.970 1.017 1.909 5 5 0.924 0.938 0.946 0.946 Best performing model Rel. RMSE Median rel. RMSE Naive model RMSE Best performing model Rel. RMSE Median rel. RMSE Notes: Fifteen-year rolling regression. RMSE is root mean squared error. Numbers in bold indicate cases in which the naive model outperforms the combination model. 40 1Q/2002, Economic Perspectives Conclusion We can summarize our main results as follows. First, we show that the A&O findings hold for a broader class of models than they studied, as well as for a longer forecasting horizon. However, they do not hold for the 197784 period. We extend their analysis to core PCE and show that the naive model does better over the sample period considered by A&O at the one-year horizon, but not at the two-year horizon. In the 1993 2000 period, the Phillips curve models perform well at forecasting core PCE for both horizons. Second, we show that Phillips curve models have predictive power for the direction of change in inflation. This is particularly true in the 197784 and 19932000 periods. However, in the 198592 period, the gain over the naive model is quite modest. Third, in most cases it is possible to combine the information in the directional forecasts with the naive model to improve on the latter models forecasts. A common thread in our results is the relatively poor performance of the Phillips curve models in the middle period, in terms of both magnitude and direction. We believe this is due to a reduction in inflation volatility and the change in monetary policy operating characteristics that took effect in this time. Our findings suggest the following policy recommendation. If we expect the current monetary regime to persist, then we can have some degree of confidence in the Phillips curve models going forward. On the other hand, if we suspect that a regime shift has recently occurred, then we should be skeptical of the Phillips curve forecasts. In any case, there may be some directional information in these forecasts, and we can use this to improve on naive forecasts. Our findings suggest that more empirical and theoretical work is necessary to come to a complete answer to the question raised in the title to this article. An equivalent way of posing this question is to ask: Why does inflation behave like a martingale over some periods while at other times it does not? We have suggested some possible explanations. Empirically, we need to assess the robustness of our results to crosscountry analysis. For example, here we have only one regime change and, hence, only one observation for the regime-switch hypothesis. Ultimately, assessing the plausibility of various possible explanations will require developing a fully specified theoretical model. Such work may shed light on the connection between monetary policy and aggregate outcomes, as well as the nature of the price-setting mechanism. NOTES 1 For a recent discussion of the intellectual history of the Phillips curve and NAIRU, see Gordon (1997). 6 2 See Bernanke and Mihov (1998), Bordo and Schwartz (1997), and Strongin (1995) for discussions of monetary regimes. These papers argue that during the Volker chairmanship of the Board of Governors from 197987 monetary policy shifted, in terms of operating procedures and the Feds increased willingness to combat inflation. Furthermore, Bernanke and Mihov (1998) estimate monetary policy rules and can reject the hypothesis of parameter stability for dates in the 1980s. 7 Suppose there are K lags in the equation, thenÿb(L)xt = b0xt + b1xt-1 + bKxt-K, where the b parameters are scalars. 9 3 4 For more details on the CFNAI, see www.chicagofed.org/ economicresearchanddata/national/index.cfm. With eight models, the median is the average forecast of the fourth and fifth ranked forecasts. 5 Federal Reserve Bank of Chicago One implication of this procedure is that the historical path of the indexes may change between forecast dates. The alternative measure we used was mean absolute value error. T 12 ˆ 12 For J = 12, 24, this is expressed as (1/ T ) i 1 Q i J Q i J . ¤ A&O use several sample periods for their analysis. When they consider unemployment-rate-based Phillips curves, their sample begins in 1959. When they consider CFNAI-based Phillips curves, their sample begins in 1967. S&W begin their analysis in 1959. 8 We also considered estimating the forecasting equations using all the data from 1967 up to the forecast date. The results obtained indicated either very similar forecast performance or, in a few cases, a slight deterioration relative to the rolling regression procedure. S&W report evidence supporting this hypothesis. They find that unemployment-rate-based Phillips curve forecasting models exhibit parameter instability during the 1980s. 10 41 REFERENCES Atkeson, Andrew, and Lee E. Ohanian, 2001, Are Phillips curves useful for forecasting inflation?, Federal Reserve Bank of Minneapolis, Quarterly Review, Vol. 25, No. 1, Winter, pp. 211. Bernanke, Ben, and Ilian Mihov, 1998, Measuring monetary policy, Quarterly Journal of Economics, Vol. 113, No. 3, August, pp. 869902. Bordo, Michael, and Anna Schwartz, 1997, Monetary policy regimes and economic performance: The historical record, National Bureau of Economic Research, working paper, No. 6201. Stock, James H., and Mark W. Watson, 2001, Forecasting output and inflation: The role of asset prices, Princeton University, manuscript, February. , 1999, Forecasting inflation, Journal of Monetary Economics, Vol. 44, pp. 293335. Strongin, Steven, 1995, The identification of monetary policy disturbances: Explaining the liquidity puzzle, Journal of Monetary Economics, Vol. 35, pp. 463497. Gordon, Robert, 1997, The time-varying NAIRU and its implications for economic policy, Journal of Economic Perspectives, Vol. 11, No. 1, Winter, pp. 1132. 42 1Q/2002, Economic Perspectives May 8-10, 2002 38th ANNUAL CONFERENCE ON BANK STRUCTURE AND COMPETITION FEDERAL RESERVE BANK OF CHICAGO Behavior over the Business I -J Os', ■ . •• . A On May 8-10, 2002, the Federal Reserve Bank of Chicago will hold its 38th annual Conference on Bank Structure and Competition at the Fairmont Hotel in Chicago. Since its inception, the conference has encouraged an ongoing dialogue and debate on current public policy issues affecting the financial services industry. Each year the conference brings together several hundred financial institution executives, academics, and domestic and international regulatory authorities to examine topical policy issues. Ihe 2002 conference will evaluate the changing condi Today, these institutions are more geographically diversified, tion and behavior of financial institutions over the business cycle and the appropriate regulatory offer a wider range of financial products, and are better able to hedge risk—all of which may make them more resilient response. While we have seen unprecedented economic against adverse economic shocks. However, it remains expansion over the past decade, recent events indicate that unclear as to whether the new financial system is actually the business cycle with its associated booms and reces sions has not been eliminated. As we move through the less susceptible to cycles. Given this uncertainty, what is the appropriate regulatory and supervisory policy over the business business cycle, there is a need to understand the forces cycle? Some argue that policy should be countercyclical: more driving financial firm behavior as well as how firms respond to changing credit conditions and market demands. It is also important to ponder the alternative policy options available to regulators across the cycle and the impact of those options on the industry. The policy debate concerning firm and supervisory/regulatory behavior over the business cycle has intensified recently with stringent during upturns and more accommodating during downturns to avoid exacerbating an already bad situation. Others argue that sound regulatory and supervisory policy regarding credit quality and bank condition must be based on established standards that are independent from the business cycle. As in past years, much of the program will be devoted to the primary conference theme, but there will also be a number of proposed modifications to the Basel Capital Accord. In the sessions on current industry issues. Some of the highlights of early 1990s, financial institutions responded to an economic the conference include: slowdown and changing capital requirements by adjusting their credit standards and, as a result, credit availability. A perceived "credit crunch" ensued, and recovery from that ■ The keynote address by Federal Reserve Board Chairman Alan Greenspan. recession was unusually slow. Deposit flows were also dis rupted, and as the economy recovered some institutions found it more difficult than others to raise core deposit funds using ■ A discussion of the theme from a variety of perspectives traditional means. As we proceed through the current slow by a panel of industry experts. Participants on this panel include Charles Goodhart, the Norman Sosnow Professor down, there is reason to believe that lending and deposit of Banking and Finance at the London School of Business; taking institutions may react differently than in the past. Sean Ryan, Fulcrum Global Partners; Karen Shaw Petrou, Managing Partner, Federal Financial Analytics; and Richard Spillenkothen, Board of Governors of the Federal Reserve System. ■ Luncheon presentations by John Hawke, Jr., Comptroller of the Currency, and Donald E. Powell, Chairman, Federal Deposit Insurance Corporation (FDIC). ■ A discussion of the events of September 11 and the impli cations for the financial services sector by Roger Ferguson, Participants include John Bovenzi, Chief Operating Officer, FDIC; Mark Harding, Partner, Clifford Chance LLP; Oliver Page, Financial Services Authority; Robert Pickel, Chief Executive Officer, International Swap Dealers Association; and Ken Scott, Stanford Law School. As usual, the Wednesday sessions (May 8th) will showcase more technical research that is of primary interest to research economists in academia and government. The Thursday and Friday sessions are designed to address the interests of a Vice Chairman, Board of Governors of the Federal Reserve broader audience. System; Michael Chertoff, Assistant Attorney General, U.S. Department of Justice; and Eileen Wallace, Managing Director, Morgan Stanley Dean Witter, Inc. If you are not currently on our mailing list or have changed ■ A panel discussion of issues related to the current bank capital reform proposal by Nicholas Le Pan, Canadian your address and would like to receive an invitation to the conference, please contact: Ms. Portia Jackson Superintendent of Financial Institutions and Head, Basel Conference on Bank Structure and Competition Capital Accord Implementation Group; Richard J. Herring, Research Department Jacob Safra Professor of International Banking, The Federal Reserve Bank of Chicago Wharton School; Michael Ong, Executive Vice President 230 South LaSalle Street and Chief Risk Officer, Credit Agricole Indosuez; and Chicago, Illinois 60604-1413 Eric Rosengren, Senior Vice President, Federal Reserve Bank of Boston. Telephone: 312-322-5775 email: portia.jackson@chi.frb.org ■ A panel discussion of the complexities involved with failure resolution of large, complex banking organizations. Origins of the use of Treasury debt in open market operations: Lessons for the present David Marshall Introduction and summary From late 1997 through the third quarter of 2001, continuing fiscal surpluses by the federal government caused the outstanding stock of Treasury debt to decrease substantially. While the onset of the current recession, along with the recent tax cuts, has slowed or even reversed this trend, many analysts believe that the declines in Treasury debt will resume over the next decade once the economy starts to strengthen. This could present an operational problem for the Federal Reserve. The Fed currently injects liquidity into the economy by expanding bank reserves via open market operations. That is, the Federal Reserve expands liquidity by purchasing securities on the open market and withdraws liquidity through open market sales of securities. Currently, all permanent transactions by the Federal Reserve open market desk use Treasury securities, and Treasury securities remain the primary medium for temporary transactions. As demand for currency and dollar-denominated bank reserves grows in the years to come, the Federal Reserve will have to acquire ever-increasing amounts of Treasuries via open market purchases. But if the total stock of such securities shrinks over the next decade or two, the Fed may find it increasingly difficult to conduct the needed transactions. The Federal Reserve would then have to consider changing its longstanding procedures for open market operations. In particular, the Fed may have to consider purchasing securities issued by non-governmental obligors.1 Is there a precedent for Federal Reserve trading in privately issued assets? How does the Federal Reserve choose the medium to use for open market operations? Has the Fed consistently chosen the safest or most liquid class of securities, or has it sought to influence the development of financial markets in its choice of open market instruments? In this article, I review the early history of open market operations, with an eye toward addressing Federal Reserve Bank of Chicago these questions. The historical record shows that prior to the U.S.’s entry into World War I, the Federal Reserve’s preferred media for open market operations were private bills of exchange, trade acceptances, and bankers’ acceptances,2 rather than public debt. The Federal Reserve’s choice was influenced by the prevailing theory of monetary policy, known as the real bills doctrine, which held that the central bank should only provide liquidity in exchange for securities that directly finance commerce. In addition, the Federal Reserve’s use of private acceptances in open market operations was in part an effort to encourage the development of an active secondary market in private paper. At the same time, the Federal Reserve was rather reluctant to hold large quantities of Treasury securities. Purchases of government debt by the central bank were seen as tantamount to “lending to the crown,” which was regarded as a dangerous path for central bank policy. Furthermore, there were problems of coordination with the Treasury that took several years to resolve. The Federal Reserve eventually moved away from private paper toward Treasury securities for several reasons. The supply of Treasuries expanded rapidly during World War I due to the financing needs of the war. Concomitantly, the secondary market in Treasuries grew rapidly. The supply of private paper contracted during the recessions of 1920–21 and (more importantly) 1929–33. Finally, events during the 1920s caused monetary theorists to become disenchanted with the real bills doctrine. David Marshall is a senior economist and economic advisor at the Federal Reserve Bank of Chicago. The author thanks Jim Clouse, Anne Marie Gonczy, Ed Green, Thomas Simpson, and François Velde for helpful comments. 45 So, what do we learn from this review of history? First, there were extended periods when the Federal Reserve conducted open market operations primarily in private securities. Furthermore, the Federal Reserve used its choice of open market instruments to influence the growth of financial markets in ways it deemed useful for the public interest. Finally, a shift to a new set of open market instruments may have unforeseen side effects. It takes time to understand the full implications of a major change in operating procedures, so a gradual transition may be the best way to proceed. In the next section, I discuss the issues confronting Federal Reserve open market operations as the stock of Treasury debt shrinks. I then describe how open market operations evolved from the earliest days of the Federal Reserve through the Great Depression. Finally, I discuss how this historical record might have relevance to the issues of the present day. The problems currently facing open market operations An important source of liquidity in the U.S. economy is the monetary base, M0, which consists of currency in circulation plus bank reserves. M0 comprises about 97 percent of Federal Reserve liabilities. These liabilities are balanced primarily by securities purchased on the open market (approximately 96 percent of Federal Reserve assets). The other main way that the Federal Reserve expands liquidity is by lending to commercial banks at the discount window. However, discount window loans represent a very small fraction (currently 0.015 percent) of Federal Reserve asset holdings. The vast majority of Federal Reserve security holdings—currently 95 percent—consist of U.S. Treasury securities. The Federal Reserve has conducted open market operations primarily in Treasury securities since the mid-1930s, and Treasuries are the only medium it has used for its outright transactions since 1981.3 Thus, to a close approximation, every dollar’s worth of M0 in circulation is matched on the Fed’s balance sheet by one dollar’s worth of U.S. Treasury securities acquired through open market purchases. This fundamental balancing relationship presents a problem: Demand for M0 is growing rapidly, while the stock of Treasury debt that the Federal Reserve uses to balance M0 has been shrinking. The black line in figure 1 plots M0 from 1986 through the present (indicated by the vertical line). Over this period, the monetary base grew at a geometric rate of around 6.8 percent per year. This is mostly due to a growing demand for currency. In 1975, currency accounted for about 77 percent of M0. Since February 2000, however, this 46 fraction has exceeded 90 percent. (The only exception occurred during the two weeks following the September 11 attacks, when the fraction of M0 represented by currency dropped to 86 percent. This was due to the Federal Reserve’s temporary expansion of bank reserves in response to the attacks.) The growth in M0 is due in part to the growth in domestic economic activity. In addition, much of the increased demand for currency is due to increased demand for dollars abroad. Consider two examples: Ecuador formally replaced the sucre with the dollar as its official currency in 2000; and, while the peso continues to be the official currency in Argentina, around 60 percent of transactions in Argentina are actually conducted with dollars. (See Velde and Veracierto, 2000.) These trends are likely to continue inducing growth in demand for the U.S. monetary base. The black dashed line in figure 1 plots a projected path for M0 through 2011.4 The projection is a mechanical extrapolation of past trends and is not intended as a detailed forecast. Nevertheless, it is a plausible first guess at how the monetary base might evolve over time. Figure 1 shows the monetary base approximately doubling in the next ten years. To accommodate this growing demand for M0, the stock of assets owned by the Federal Reserve must grow. If the Federal Reserve continues its current policy of maintaining virtually all its asset holdings in the form of Treasury securities, its ownership of Treasury debt will have to expand rapidly. However, the total quantity of Treasury securities may well fall during the coming years. The green line in figure 1 plots the stock of outstanding Treasury debt from 1986 to the present. Note that the level of Treasury debt had fallen from $3.8 trillion in November 1997 to $3.3 trillion as of September 2001, a decrease of over 10 percent in less than four years. In spite of the recession that started in March 2001 and the 2001 tax cuts, the contraction in Treasury debt continued at least through the third quarter of the year. The green dashed line in figure 1 plots the path of Treasury debt implied by the Congressional Budget Office’s (CBO) most recently published forecasts of federal surpluses through 2012, released in January 2002.5 These forecasts take the effects of the current recession into consideration. The CBO predicts small deficits through early 2004, followed by surpluses. The problem facing the Federal Reserve can be seen by comparing the two forecasts in figure 1. Taking these forecasts at face value, the stock of base money demanded by the economy would equal the stock of Treasury debt in July 2012. This means that the 1Q/2002, Economic Perspectives FIGURE 1 Monetary base versus Treasury debt billions of dollars 4,500 Treasury debt 4,000 3,500 3,000 2,500 2,000 1,500 1,000 Monetary base 500 0 1986 '89 '92 '95 '98 '01 '04 '07 '10 Note: Dashed lines indicate projected paths. Source: U.S. Treasury, Federal Reserve Board, and author’s calculations. Federal Reserve could not accommodate the growing demand for M0 beyond that date without purchasing securities other than Treasuries. In fact, the problem will arrive much sooner. The Federal Reserve recognizes that Treasury securities serve a unique role in financial markets. Because they are free of default risk and are highly liquid relative to other assets,6 Treasuries are a preferred savings instrument for foreign investors and are extensively used for hedging and as benchmarks for pricing other fixed-income securities. If the Federal Reserve held a large fraction of outstanding Treasury securities, it would impair the liquidity of Treasury markets, which could adversely affect other markets and even affect the pace of economic activity. As a result, the Federal Reserve limits its fraction of ownership of any individual Treasury issue. The current ownership caps range from 35 percent for Treasury securities with less than a year to maturity to 15 percent for issues ten years and longer. If the Federal Reserve continues to abide by these caps, it will exhaust its capacity to acquire additional Treasury securities long before the outstanding stock of Treasury debt disappears. Possible responses to these issues Clearly, the Federal Reserve may well have to modify its current procedures for conducting open market operations. The Fed could relax its self-imposed Federal Reserve Bank of Chicago caps on Treasury holdings, but this could impair liquidity in the Treasury market and, in any event, would only represent a temporary stopgap. A longer-run solution would be for the Federal Reserve to start including in its portfolio assets other than Treasuries. Under current law, the Federal Reserve can purchase a range of assets, including direct obligations of federal agencies or debt fully guaranteed by federal agencies, debt of foreign governments, certain state and local obligations, and selected other instruments. There has been speculation in the press that the Federal Reserve might seek legislation to expand its authority to hold private assets.7 The extension of the Federal Reserve’s portfolio to non-Treasury securities raises a number of questions. Which assets should '13 the Federal Reserve hold? Should it let private market participants align on a new substitute for Treasury securities and then simply adopt this asset class? Alternatively, should the Federal Reserve actively seek to influence the evolution of fixed-income markets as they adjust to an era of diminishing supply of Treasury securities? In particular, should the Federal Reserve attempt to steer the market toward the type of Treasury substitute that it prefers? The sorts of choices the Federal Reserve now faces are not unprecedented. In the following sections, I review the early history of open market operations. This account shows that, in the early days of the Fed, Treasury securities were not the preferred medium for open market operations. Only gradually did Treasuries displace other assets. Furthermore, the Federal Reserve’s original intentions for open market operations included a desire to affect the evolution of financial markets. In particular, it sought to encourage an active secondary market in acceptances. Thus, there are antecedents both for the Federal Reserve holding privately issued securities and for the Federal Reserve using its open market procedures to influence the development of financial markets. Having said this, financial markets have changed enormously since the early years of the Federal Reserve System, so we should use caution in drawing lessons from these precedents for current problems. The early years of open market operations The current practice of conducting Federal Reserve System open market operations almost exclusively with Treasury securities was not anticipated in 47 the earliest years of the Fed. Table 1, taken from Meulendyke (1998), shows that bankers’ acceptances were the primary asset class for the Federal Reserve portfolio until World War I, and acceptances had a roughly equal presence with Treasury securities through the 1920s. Treasury securities did not become predominant until the Great Depression. These patterns reflect changes both in the thinking of Federal Reserve officials and in the economic environment in which the Fed operated. At the inception of the Federal Reserve in 1913, it was presumed that Federal Reserve assets would primarily consist of short-term privately issued paper, such as bankers’ acceptances, trade acceptances, and bills of exchange.8 A key reason for this focus was the real bills doctrine, which was the most influential theory of central banking at the beginning of the twentieth century. The real bills doctrine maintains that “a banking system that confines its lending to discounting short-term self-liquidating commercial bills of exchange arising from real transactions in goods and services—the productive use as opposed to the speculative use of credit—cannot over-issue.”9 That is, the banking system would not create excessive (and therefore inflationary) amounts of credit. A particularly important exponent of this view was Paul Warburg, TABLE 1 Federal Reserve holdings, 1915–50 Year-end Treasury securities (dollars in millions) 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 16.0 55.0 122.0 238.0 300.5 287.4 234.1 433.4 133.6 540.2 374.6 314.8 560.0 197.2 487.3 686.1 774.6 1,851.1 2,435.3 2,430.3 2,430.3 2,430.2 2,564.0 2,564.0 2,484.2 2,184.1 2,254.5 6,188.7 11,543.0 18,846.1 24,262.3 23,349.7 22,559.4 23,332.8 18,884.6 20,724.5 Bankers’ acceptances (percent of total) 19.8 31.2 31.4 45.5 80.8 60.6 61.8 61.5 27.5 58.3 50.2 45.2 64.4 31.1 67.4 70.4 78.3 99.8 95.7 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 (dollars in millions) 64.8 121.2 266.9 285.3 71.6 187.2 145.0 271.0 352.0 386.9 372.2 381.0 308.9 437.5 235.3 288.8 215.3 3.6 108.1 0.1 0.0 0.0 0.5 0.5 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 (percent of total) 80.2 68.8 68.6 54.5 19.2 39.4 38.2 38.5 72.5 41.7 49.8 54.8 35.6 68.9 32.6 29.6 21.7 0.2 4.3 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Source: Meulendyke (1998, table 1, p. 22). 48 1Q/2002, Economic Perspectives “[T]he acquisition of Government securities by a banker with Kuhn, Loeb & Co., whose pamphlet, the central bank is regarded as opening the door “A plan for a modified central bank” (published durto inflation. It is usual for the power of the central ing the Financial Panic of 1907), strongly influenced bank to lend to the Government to be carefully the drafting of the Federal Reserve Act of 1913. circumscribed, and the dividing line between lendWhile the real bills doctrine does not distinguish ing direct and buying Government securities in between bills acquired through rediscounting and bills the market may be rather a fine one.” acquired through open market purchases, there was a Similarly, in his pamphlet, “Principles that must underperception that open market transactions by the Fedlie monetary reform in the United States” (published eral Reserve in the commercial bills market would be in November 1910), Warburg warns against the inflabeneficial to the economy. In particular, Warburg and tionary danger of issuing notes backed by government others believed that active trading of real bills by the bonds (Warburg, 1930b, p. 176). Fed could help foster the development of a secondary In addition, it was thought that a well-run central market in these securities. Unlike most Western bank should be free from political influence.15 ExtenEuropean countries, the U.S. did not have an active sive holdings of government debt might compromise market in acceptances prior to the passage of the Fedcentral bank independence.16 Finally, the real bills doceral Reserve Act. Warburg and others saw the develtrine emphasized that central bank assets should be short opment of a liquid acceptance market as essential for term. At the time the Federal Reserve was established, a modern banking system to emerge in the U.S. In there were no short-term Treasury bills, and there was Warburg’s words, “We should aim to transform our no active market in short-term Treasury securities.17 commercial paper from a non-liquid asset into the 10 Thus, Federal Reserve assets prior to World War I quickest [that is, most liquid] asset of our banks.” were primarily short-term commercial bills. Table 3, Prior to the establishment of the Federal Reserve, the taken from Reynolds (1922b, p. 77) shows that until most liquid short-term assets traded were call loans mid-1917, commercial bills purchased in the open marused to finance stock purchases on the New York Stock ket accounted for a larger fraction of Federal Reserve Exchange. This form of investment was seen as specassets than government securities. From the beginning ulative, rather than productive. It was hoped that the of 1916 through the start of World War I, bills purdevelopment of a liquid market in real bills would facilitate the optimal allocation of credit across induschased in the open market exceeded those acquired tries and geographical regions11 and divert credit through rediscounting.18 toward productive investment and away from more World War I and its aftermath speculative uses.12 This development would be fosIt was America’s entry into World War I in April tered by a central bank that actively purchased these 1917 that spurred the big shift away from this focus bills in the open market.13 Pursuant to this goal, the on commercial bills. The war was largely funded by System’s purchases of commercial bills (primarily government debt. The Federal Reserve was reluctant trade and bankers’ acceptances) in the open market vastly exceeded its acquisition of these securities via to buy government debt directly from the Treasury,19 rediscounting. We can see this in table 2, which is taken from Agger (1922, table 2, TABLE 2 p. 216).14 Federal Reserve Bank discounts and purchases In contrast to the focus on real bills of trade and bankers’ acceptances (particularly bankers’ acceptances) as ap(in thousands of dollars) propriate assets for a central bank, there was some concern about central bank purPurchases in Discounts open market chases of government securities. Direct Year Bankers’ Trade Bankers’ Trade loans to the government were seen as dangerous, tying the supply of credit to the 1915 1,959 64,814 31 1916 5,212 369,762 16,333 spending whims of the government. Open 1917 37,771 1,046,765 30,948 market purchases of government securities 1918 19,940 187,373 1,748,503 61,036 were seen as equivalent to this sort of di1919 71,643 138,420 2,788,619 36,558 rect lending. Furthermore, monetizing 1920 187,162 192,157 3,143,737 74,627 government debt was seen as inflationary. 1921 (9 months) 49,810 101,129 996,851 6,687 Hawtrey (1933, p. 131) summarized these Source: Agger (1922, table 2, p. 216). ideas as follows: Federal Reserve Bank of Chicago 49 but, under government pressure, it took steps to accommodate the increased supply of Treasury securities. It accepted Treasury securities from member banks for rediscounting; it accepted bills backed by Treasuries from member banks for rediscounting; and it offered a lower rate on loans collateralized with Treasury securities than with other forms of collateral.20 As a result of these steps, the Federal Reserve’s portfolio became heavily based (directly or indirectly) on Treasury debt. By May 1919, 95.2 percent of Fed purchases of commercial bills (total of rediscounts plus open market purchases) were backed by government securities. Open market purchases of government securities also increased dramatically. According to West (1977), such purchases amounted to only $4.37 million in April 1917. By March 1918, purchases of government securities amounted to $1,099 million (55.1 percent of total investments).21 Once the war ended, purchases of government securities suffered a decline relative to acceptances. However, the war did result in permanent changes in Federal Reserve operations. First, the war years established a precedent for extensive Federal Reserve holdings of government debt. However, there were those who advocated withdrawing from the Treasury market following the end of the war. For example, Welton and Crennan (1922) attributed the inflation during World War I to the backing of currency with government securities, and argued that this practice should stop. Second, the volume of Treasury securities issued to finance the war created an active market in government debt.22 A third development was a growing disenchantment with the real bills doctrine. In particular, the recession of 1920–21 was caused in part by excessive inventory building. The inflation following the end of the war motivated firms to hold speculative inventories, hoping to sell at higher prices. These inventories were financed, in part, by commercial bills, which were suitable assets for rediscounting under the real bills doctrine. The central bank credit thus created further fueled the inflation. As a result, Federal Reserve officials (notably Benjamin Strong, president of the Federal Reserve Bank of New York) argued that the real bills doctrine was neither necessary nor sufficient to avoid inflationary credit expansion or to ensure that credit would be used for productive, rather than speculative activity.23, 24 These developments weakened the Federal Reserve’s original focus on real bills. Thereafter, purchases of real bills (mostly bankers’ acceptances) and government securities coexisted. The recession of 1920–21 severely contracted the supply of bankers’ acceptances and other real bills.25 The Fed responded by replenishing its earning assets with government debt.26 However, it appears that real bills were used for secular growth of the Federal Reserve’s open market portfolio, while government securities were used to TABLE 3 Earning assets of the Federal Reserve System 1914–17 (in thousands of dollars) Date 1914 December 31 Bills discounted for members Bills bought in open market 9,909 U.S. government securities Municipal warrants 205 734 1915 January 29 April 30 July 30 October 29 13,955 22,774 29,102 30,448 13,812 11,625 13,619 2,015 6,813 7,923 10,505 11,165 18,656 16,107 25,014 1916 January 28 April 28 July 28 October 27 26,901 21,448 27,594 21,131 26,314 47,585 83,454 86,085 21,372 45,841 48,656 40,469 20,602 36,933 27,220 29,890 1917 January 26 April 27 July 27 October 26 15,711 35,043 138,459 397,094 97,697 71,400 195,097 177,590 55,769 117,818 76,953 110,042 12,249 14,999 1,469 233 Source: Reynolds (1922b, p. 77). 50 1Q/2002, Economic Perspectives manage aggregate credit provision in the FIGURE 2 short term. In particular, figure 2 (taken Open market paper and government from chart 9.3 in West, 1977, p. 191) securities purchased, 1921–23 shows that the variability of government thousands of dollars securities purchases from 1921–23 was 2,000 much greater than the variability of open 27 market purchases of real bills. This pattern suggests that Treasury securities 1,500 served a role analogous to present-day temporary transactions. In the early 1920s, an argument against extensive open market operations 1,000 Government in government securities came from the securities Treasury itself. The Reserve Banks’ iniOpen market tial open market activity in government paper 500 debt was uncoordinated, causing random fluctuation in the pricing of these securities. This presented a problem for the 0 Treasury, making it more difficult to foreJan May Sept Jan May Sept Jan May Sept cast auction prices.28 In part to address the 1921 ‘22 ‘23 Treasury’s concerns, in May 1922 the Conference of Presidents of the Federal Source: West (1977, chart 9.3, p. 191). Reserve Banks established a committee on centralized execution of purchases and sales of government securities to coordinate all acceptances simply infeasible. Evidence on this point Federal Reserve purchases of Treasury securities.29 is provided by Groseclose (1965, p. 132): “[D]uring In 1923, this committee was reconstituted under the the boom just preceding the stock market crash of supervision of the Board of Governors as the Open 1929 the volume of bankers’ acceptances rose to around Market Investment Committee, the precursor of the $1.5 billion, but thereafter declined to less than $150 current Federal Open Market Committee. The Treasury’s million at the end of 1941. ...” As a result, “[a]fter 1937 concerns appeared to restrain the growth in the use of the Federal Reserve practically ceased to buy or redisgovernment debt for open market operations. count such paper” until after World War II. At the same time, the government started issuing short-term debt The Depression, 1929–33 on a regular basis. The first Treasury bill issue was in The event that ultimately caused a permanent shift December 1929,32 providing the Fed with an alternaaway from bankers’ acceptances to government secutive to bankers’ acceptances as a short-term instrument rities was the Great Depression. According to Anderson for open market operations. (1965), there was a consensus that the Federal Reserve’s A final impetus for extensive Federal Reserve response to the market crash in fall 1929 should include holdings of government debt was provided by the aggressive open market purchases. Anderson wrote Roosevelt administration’s national recovery actions that, “Acceptances and, if necessary, [italics added] in 1933. As with the costs of World War I, the governgovernment securities should be purchased to avoid ment financed these actions with debt. The Treasury any increase and possibly to bring some reduction in needed to ensure that debt issues were successful, and member-bank indebtedness to the Reserve Banks.”30 the Federal Reserve responded to the Treasury’s conInterestingly, this quote suggests that acceptances, cerns. According to Anderson (1965, p. 72), “There rather than government securities, were seen as the was a consensus that with excess reserves still substanprimary vehicle for increasing bank liquidity. tial, it was not desirable to buy government securities After 1929, however, it is difficult to find any mento increase bank reserves. ... There was apprehension tion of acceptances in discussions of open market op[however] that if the Treasury could not do its financerations.31 It appears that the aggregate supply of ing successfully in the market, it would be forced to acceptances fell with the decline in economic activiseek accommodation directly from the Reserve Banks.” ty, rendering extensive open market operations in As a result, in spring 1933 the Board of Governors Federal Reserve Bank of Chicago 51 authorized the purchase of up to $1 billion of government securities, if necessary, to ensure successful financing by the Treasury. What can we learn from the historical record? We must be cautious in drawing lessons from this historical account of open market operations. For one thing, the events described in the preceding sections all occurred under the gold standard, a very different monetary environment from the present. In addition, current financial markets are far more highly developed than in the early years of the Fed. Nonetheless, there are a number of parallels between the System’s experiences in its early years of existence and the policy choices that the System may face over the next few years. During the first three decades of the last century, the Fed went through the process of changing the class of securities used in open market operations. The problem encountered by the Federal Reserve during the recession of 1920–21 and the Great Depression resembles that currently facing the System: a dwindling supply of the assets traditionally used for open market operations. In the 1930s, the problem was a reduced supply of acceptances induced by the economic contraction, while currently it is the possibility of a reduced supply of Treasuries. The adjustment in the 1930s to a Treasuries-only policy was not immediate. It took many years for the System to reconcile the advantages of using Treasury securities with their associated problems, most notably the problem of central bank independence highlighted by Bagehot (1873) some 70 years before. Ultimately, these issues were not to be fully resolved until the Treasury–Federal Reserve Accord of 1951, in which the Treasury agreed that the Fed should be permitted to pursue an independent monetary policy. In its early years, the Fed used open market operations to affect the development of private markets. Specifically, the System deliberately used the purchase of private bills in the open market to foster the 52 development of a liquid secondary market in acceptances. This action stands in contrast to the Federal Reserve’s current policy of minimizing market distortions, wherever possible, in its open market activities. In the early days of the System, however, concerns about creating distortions in financial markets were outweighed by other public policy considerations. At that time, financial markets and the banking system were not well developed. The Federal Reserve’s activities might therefore be seen as serving a public policy purpose by addressing a market incompletion. Today’s markets are so much more highly developed that it is difficult to make a case for this sort of active interventionist policy. Nonetheless, the Federal Reserve still faces a basic issue that was recognized in its early years: Its role in financial markets may have an influence on market outcomes. If the Fed moves toward accepting privately issued securities in its open market account, this policy shift may affect the evolution of markets. For example, as the supply of Treasury securities contracts, private markets will align on some alternative benchmark security to replace the ever-scarcer Treasuries. The System’s choice of private assets to use in its open market operations may influence the class of securities that emerges as the new benchmark. In addition, if the Fed purchases private securities, it might be seen as selectively approving those obligors whose paper it purchases. When the Fed discontinued all purchases of acceptances in 1984 (it discontinued outright purchases of acceptances in 1977), this concern was a major factor. In the words of President Solomon of the Federal Reserve Bank of New York, “There are some people ... who misinterpret the Federal Reserve eligibility as a good housekeeping seal.”33 While there are antecedents for open market operations in private securities, there clearly are fundamental problems that must be addressed should the Federal Reserve consider using private securities in this way in the future. 1Q/2002, Economic Perspectives NOTES 1 Broaddus and Goodfriend (2001) propose that the Treasury continue issuing bonds sufficient to meet the Federal Reserve’s needs, purchasing private assets with the proceeds if necessary. This would transfer the responsibility of holding private assets from the Federal Reserve to the Treasury. A bill of exchange is a negotiable security issued by one party (the “drawer”) and accepted by the other party (the “drawee”), instructing the drawee to pay a fixed sum of money, usually as part of a commercial transaction. It differs from a promissory note only in that it is initiated as an instruction from the creditor, rather than as a promise from the debtor. A trade acceptance is essentially a bill of exchange issued in the course of an export/import transaction. It is an obligation of the buyer in the transaction. A bankers’ acceptance is a trade acceptance that has been guaranteed by the buyer’s bank, at which point it becomes an obligation of the bank, rather than of the buyer. 2 15 See, for example, Bagehot (1873), chapter 4. 16 Warburg, (1930b), p. 172. 17 Warburg, (1930b), p. 169. 18 In comparing tables 2 and 3, note that table 2 gives cumulative purchases over the year, while table 3 gives point-in-time asset stocks. 19 The Reserve Banks did agree to take a $50 million issue of 90day certificates of indebtedness. (See West, 1977, p. 187.) 20 See Reynolds (1922a), p. 191, and West (1977), pp. 187–188. 21 See West (1977), p. 188. 22 See West (1977), p. 192. 3 Outright transactions are purchases or sales of securities that are intended to be permanent. The Federal Reserve generally conducts outright transactions only a few times each year. In contrast, temporary transactions are purchases or sales that are expected to be reversed in the near term. Temporary transactions are conducted more frequently. For a discussion of the difference between outright (or permanent) and temporary open market transactions, see Meulendyke (1998). 4 This forecast uses a statistical model that fits a twelfth-order autoregression in the change in the log of M0. I use weekly data from January 1962 through November 2001. 5 U.S. Congress, CBO (2002). Also see the CBO website, www.cbo.gov. 6 7 8 9 See the discussion in Reinhart and Sack (2000). Temple-Raston and Weisman (2001). See Reynolds (1922b), pp. 74–75. Bordo and Schwartz (2000). 23 See the discussion in West (1977), pp. 195–201. 24 While the System moved away from the real bills doctrine during the early 1920s, the ideal of a self-regulating monetary policy has received renewed attention in recent years. Most notably, Sargent and Wallace (1982) formalize the notion of an “elastic currency” (in the terminology of the Federal Reserve Act of 1913). They show how a theoretical version of the real bills doctrine can allow both the quantity of money and the price level to respond optimally to fluctuations in real economic activity. 25 “From an estimated maximum of around $1 billion in acceptances outstanding at the height of their use [prior to the recession], the volume dropped to around $400 million in 1923. Much of this drop, of course, was due to the business recession” (Groseclose, 1965). 26 See Anderson (1965). 27 See the discussion in West (1977), p.191. 28 See Anderson (1965), p. 144. 29 See Anderson (1965), p. 51. 10 From “A plan for a modified central bank,” quoted in Warburg (1930a), p. 23. 11 See Warburg (1930a), p. 17, and Agger (1922), p. 209. 12 See West (1977), p. 185. 13 According to Reynolds (1922b), one of the key goals of the Federal Reserve System in its first two years was “to endeavor to regulate the interest rates and equalize the demand for money by the purchase of bills and acceptances in the open market” (Reynolds, 1922b, pp. 74–75, italics added). Note that in this quote the term “bills” clearly refers to bills of exchange, as Treasury bills were not introduced until 1929. 30 Anderson (1965), p. 61. 31 For example, Anderson’s (1965) extensive discussion of the debates over Federal Reserve open market policy in the 1930s focuses exclusively on government securities. 32 Bannon (1953). 33 Transcript of the FOMC meeting of March 26–27, 1984, available on the Board of Governors of the Federal Reserve System website at www.federalreserve.gov/fomc/transcripts/ transcripts_1984.htm. 14 West (1977), pp. 185–186, also notes Benjamin Strong’s efforts at the Federal Reserve Bank of New York to create an open market in commercial paper. Federal Reserve Bank of Chicago 53 REFERENCES Agger, E. E., 1922, “The development of an open market for commercial paper,” in The Federal Reserve System—Its Purpose and Work, A. D. Welton and C. H. Crennan (eds.), Annals of the American Academy of Political and Social Science, Vol. 99, January. , 1922b, “Early functioning of the Federal Reserve System,” in The Federal Reserve System— Its Purpose and Work, A. D. Welton and C. H. Crennan (eds.), Annals of the American Academy of Political and Social Science, Vol. 99, January. Anderson, Clay, 1965, A Half-Century of Federal Reserve Policymaking, Philadelphia: Federal Reserve Bank of Philadelphia. Sargent, Thomas J., and Neil Wallace, 1982, “The real bills doctrine versus the quantity theory: A reconsideration,” Journal of Political Economy, Vol. 90, pp. 1212–1236. Bagehot, Walter, 1873, Lombard Street, London: John Murray. Bannon, Richard J., 1953, “History of the weekly combined statement of the twelve Federal Reserve Banks,” Catholic University of America, doctoral dissertation, May. Bordo, Michael D., and Anna J. Schwartz, 2000, “The performance and stability of banking systems under ‘self-regulation’: Theory and evidence,” The Cato Journal, Vol. 14, No. 3. Broaddus, J. Alfred, and Marvin Goodfriend, 2001, “What assets should the Federal Reserve buy?,” Federal Reserve Bank of Richmond, Economic Quarterly, Vol. 87, No. 1, Winter, pp. 7–22. Groseclose, Elgin, 1965, Fifty Years of Managed Money, London: Macmillan and Company, Ltd. Hawtrey, R. G., 1933, The Art of Central Banking, London: Longmans, Green and Co. Meulendyke, Ann-Marie, 1998, U.S. Monetary Policy and Financial Markets, New York: Federal Reserve Bank of New York. Reinhart, Vincent, and Brian Sack, 2000, “The economic consequences of disappearing government debt,” Brookings Papers on Economic Activity, Washington, DC, No. 2, pp. 163–209. Reynolds, George, M., 1922a, “Rediscount rates, bank rates and business activity,” in The Federal Reserve System—Its Purpose and Work, A. D. Welton and C. H. Crennan (eds.), Annals of the American Academy of Political and Social Science, Vol. 99, January. 54 Temple-Raston, Dina, and Jonathan Weisman, 2001, “Fed may purchase private bonds,” USA Today, September 6, p. 1A. U.S. Congress, Congressional Budget Office, 2002, The Budget and Economic Outlook, Fiscal Years 2003–12, Washington DC: U.S. Government Printing Office, forthcoming, January 31. Velde, François, and Marcelo Veracierto, 2000, “Dollarization in Argentina,” Federal Reserve Bank of Chicago, Economic Perspectives, First Quarter, pp. 24–36. Warburg, Paul M., 1930a, The Federal Reserve System; Its Origin and Growth, Vol. I, New York: The MacMillan Company. , 1930b, The Federal Reserve System; Its Origin and Growth, Vol. II: Addresses and Essays, 1907–1924, New York: The MacMillan Company. Welton, A. D., and C. H. Crennan, 1922, “The integrity of the Federal Reserve System,” in The Federal Reserve System—Its Purpose and Work, A. D. Welton and C. H. Crennan (eds.), Annals of the American Academy of Political and Social Science, Vol. 99, January. West, Robert Craig, 1977, Banking Reform and the Federal Reserve, 1863–1923, Ithaca, NY: Cornell University Press. 1Q/2002, Economic Perspectives