The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Federal Reserve Bank of Philadelphia MARCH • APRIL 1990 Why Are So Many New Stock Issues Underpriced? Saunders Is There \ a Natural Rate of Unemployment?, William W. Lang MARCH/APRIL 1990 WHY ARE SO MANY NEW STOCK ISSUES UNDERPRICED? Anthony Saunders The BUSINESS REVIEW is published by the Department of Research six times a year. It is edited by Patricia Egner. Artwork is designed and produced by D ianne H allow ell under the direction of Ronald B. Williams. The views expressed here are not necessarily those of this Reserve Bank or of the Federal Reserve System. SUBSCRIPTIONS. Single-copy subscriptions for individuals are available without charge. Insti tutional subscribers may order up to 5 copies. BACK ISSUES. Back issues are available free of charge, but quantities are limited: educators may order up to 50 copies by submitting requests on institutional letterhead; other orders are limited to 1 copy per request. Microform copies are available for purchase from University Microfilms, 300 N. Zeeb Road, Ann Arbor, MI 48106. REPRODUCTION. Permission must be ob tained to reprint portions of articles or whole ar ticles. Permission to photocopy is unrestricted. Please send subscription orders, back orders, changes of address, and requests to reprint to the Federal Reserve Bank of Philadelphia, Department of Research, Publications Desk, Ten Independence Mall, Philadelphia, PA 19106-1574, or telephone (215) 574-6428. Please direct editorial communi cations to the same address, or telephone (215) 5743805. 2 According to studies, the prices of new stock issues are, on average, set below the prices investors appear willing to pay when the stocks start trading in the secon dary market. Financial economists offer various explanations for this underpricing, including the “monopoly power" of under writers, legal penalties for misinforming in a prospectus, and the costliness of col lecting information about the issuing firms. What do these explanations—and the em pirical evidence— imply for regulation of commercial and investment banks? IS THERE A NATURAL RATE OF UNEMPLOYMENT? William W. Lang In the early 1980s, unemployment rates in Europe and the U.S. reached their high est levels since the Great Depression. Since then, U.S. rates have dropped to more normal levels, but European rates are still relatively high. Citing the European experience, some economists are aban doning the idea that unemployment gravi tates toward some natural rate. They advocate a theory of "hysteresis," which argues that aggregate-demand policies can raise the unemployment rate permanently. Will theories of hysteresis replace the natural rate theory? FEDERAL RESERVE BANK OF PHILADELPHIA Why Are So Many New Stock Issues Underpriced? Anthony Saunders* Each year hundreds of small firms approach the capital market to issue equity for the first time. These firms are usually growing so fast, or have so many profitable investment projects available to them, that traditional sources of funds (bank loans, retained earnings, and the owners' own equity) are often insufficient to finance their expansion. Because of this need for finance at a crucial stage in their growth, it is important for these firms that the prices of their shares reflect the *Anthony Saunders is a Professor of Finance at New York University's Stern School of Business. He wrote this article while he was a Research Adviser to the Federal Reserve Bank of Philadelphia. true value of company assets or growth oppor tunities. In particular, if their shares are sold too cheaply, these firms will have raised less capital than was warranted by the intrinsic values of their assets. In other words, their shares will have been "underpriced." Considerable evidence shows that new or initial public equity offerings (IPOs) are underpriced on average. That is, the prices of firms' shares offered to the public for the first time are, on average, set below the prices investors appear willing to pay when the stocks start trading in the secondary market. That is, in the parlance of investment bankers, small firms appear to leave behind considerable "money on the table" at the time of a new issue. 3 BUSINESS REVIEW Why small firms raise fewer funds in the new-issue process than the market indicates they should is a crucial public policy issue. Clearly, some degree of market imperfection or lack of competition could cause such an outcome. For example, if, by restricting com mercial banks' participation in the market, the Glass-Steagall Act of 1933 has allowed invest ment bankers to enjoy a type of monopoly (market) power over new equity-issuing firms, then this would suffice to explain underpricing. Alternatively, underpricing may be the pre mium the issuing firm must pay for having little information about itself to offer potential investors. In that case, underpricing would have little to do with the regulatory structure of the investment banking industry. Let's examine the reasons for IPO underpricing and evaluate the degree to which un derpricing is due to Glass-Steagall restrictions. What is the evidence on the degree of underpricing of U.S. IPOs? What are the various explanations for underpricing? And what are the implications of these explanations, and of the associated empirical evidence, for com mercial and investment bank regulation? EVIDENCE ON UNDERPRICING In "firm commitment" underwriting ("firm" in that the investment banker guarantees the price), an investment banker (and his syndi cate) will undertake to buy the whole new issue of a firm at one price (the bid price, or BP) and seek to resell the issue to outside investors at another price (the offer price, or OP). In doing so, the investment banker offers a valuable risk-management service to the issuing firm by guaranteeing to purchase 100 percent of the new issue at the bid price (BP). The return for the investment banker in bearing underwriting risk— that is, the risk that investors will de mand less than 100 percent of the issue when it is reoffered for sale to the market— is the spread between the public offer price and the bid price (OP - BP) plus fees and commissions. (Here, 4 MARCH/APRIL 1990 and throughout this article, the term "inves tor" refers to those who buy shares through the investment banker at the offer price.) Thus, the investment banker's spread plus fees and commissions may be viewed as the direct cost of going public. However, there is also potentially an indirect cost of going public, measured by the degree to which the issue is underpriced. For example, if the BP is $5 per share and the OP is $5.25 per share, then the underwriter's spread is 25 cents per share. However, suppose that on the first day of trading in the secondary market the share price (P) closes at $7 per share. This indicates that the share has been underpriced in the new-issue process and that, potentially, the firm might have raised as much as $7 per share had it been priced "correctly." This implies that the issuing firm has borne an additional indirect new-issue cost of $1.75 per share ($7.00 - $5.25), because the investment banker has set the offer price below the price the market was willing to pay on the first day of trading. Thus, more formally, the "raw" percentage degree of underpricing (UP) of an IPO can be defined as: (1) UP = [(P - OP) / OP] x 100 where: OP = offer price of the IPO P = price observed at the end of either the first trading day, week, or month If UP is positive, the issue has been underpriced; if UP is zero, the issue is accurately priced; and if UP is negative, it has been over priced. The expression for UP is also the ex pression for a percentage rate of return. Thus, equation (1) can be viewed as the one-day (or one-week or one-month) initial return on buy ing an IPO (that is, UP = R, the initial return on the stock). FEDERAL RESERVE BANK OF PHILADELPHIA Why Are So Many New Stock Issues Underpriced? Returns calculated by equation (1) are deemed raw returns. However, researchers also com pute excess (market-adjusted) returns, as well. The reasons for this are easy to see. Given a lag between the setting of the offer price and the beginning of trading on an exchange (any where from one day to two weeks or more), the price observed in the market on the first day of trading may be high (low) relative to the offer price simply because the stock market as a whole has risen (fallen) over this period. Thus, in analyzing underpricing, reseachers need to control for the performance of the stock market in general. More specifically: (2) Rm = [(^ - 10) / I0] x 100 where: Rm = return on the market portfolio ^ = level of the general market share index at the time of listing (first day, first week, or first month) I0 = level of the market share index at the time offer is announced If Rmis positive, the market has been going up in the time between the setting of the offer price and the listing of the stock on the stock exchange. If Rm is negative, the market has been falling. Excess market or risk-adjusted initial returns (EX) can therefore be defined as:1 1 For a detailed discussion of excess returns, see Robert Schweitzer, "How Do Stock Returns React to Special Events?" this Business Review (July/August 1989) pp. 17-29. For IPOs, researchers adjust the initial return on the stock by deducting the return on the market. This is equivalent to assuming that a new IPO's returns move exactly with the market's. That is, they have a unit degree of systematic risk (or their (3 is 1). The reason for this assumption is that since IPOs have no past history of returns, one cannot estimate directly the IPO's (3 at the time of issue. The only researcher Anthony Saunders (3) EX = R - Rm According to equation (3), underpricing oc curs only when R is greater than Rm. The findings of 22 studies that examine the degree of underpricing are summarized in the table on p. 10. Although the time periods, sample sizes, and ways of calculating initial returns (especially raw versus market-adjusted) differ widely across these studies, each finds underpricing on average. For example, studies that use a one-week period to calculate the difference between the offer price and the market price of an IPO find underpricing ranging from 5.9 percent to as much as 48.4 percent.2* Thus, an important empirical fact is that U.S. IPOs are underpriced on average, result ing in small firms raising less capital than is justified by the markets' ex post valuation of their shares. WHY ARE NEW ISSUES UNDERPRICED? Several reasons have been proposed in the institutional, finance, and economics literature as to why underpricing occurs. Although this article will not discuss all the proposed rea sons, it concentrates on four views that have received much publicity. The first view attrib utes underpricing to "monopoly power" en joyed by investment bankers. The second re gards Securities and Exchange Commission regulations as the primary cause. And the third and fourth see underpricing as a problem of imperfect information among contracting parties—especially between investors and is suers. who has tried to address this problem was Ibbotson (1975), who developed an ingenious method of constructing syn thetic P’s for IPOs. 2 It should be noted that these are one week's returns and are thus very large. These underpricing "costs" swamp the direct costs of a new issue, which are, on average, in the range of 2 to 5 percent of the issue's dollar size. 5 BUSINESS REVIEW The Monopoly Power of Underwriters. One possible explanation for pervasive underpricing is the monopoly power the investment banker enjoys over the issuer.3 Given that commercial banks are barred from entering into corporate equity underwriting (a result of the GlassSteagall Act, which effectively separated com mercial banking from investment banking), investment bankers may have a degree of monopoly power that they use to earn "rents" by underpricing new issues. Of course, com petition among investment banks would limit the extent of this monopoly power. But how real is this monopoly power? Compared to U.S. commercial banks, U.S. noncommercial banking firms and foreign banks have always faced fewer restrictions on entry into investment banking. Moreover, thrifts also can enter investment banking. In recent years, for example, nonbank firms such as General Electric and Prudential have entered the investment banking industry via acquisi tions, as has Franklin Savings Bank, a thrift. This potential competition presumably places a limit on the degree of monopoly power en joyed by investment bankers. In addition, foreign banks were not subject to Glass-Steagall regulations until passage of the International Banking Act of 1978. Even then, those already possessing investment banking powers had them grandfathered. The emphasis on investment banks is due to their traditional dominance of the underwriting market and to their potential economies of scope (cost savings from offering a combina tion of services) in extending to their under writing customers a broader range of financial services. If investment bankers have monopoly power 3For a discussion of the reasons for and effects of invest ment bankers' potential monopoly power, see Ibbotson (1975) and Pugel and White (1984). 6 FRASER Digitized for MARCH/APRIL 1990 over the new issuer, they might use it to in crease both the spread between the offer price and bid price (the underwriters' spread) as well as the degree to which the offer price is set below the markets' true valuation (P). A monopolist investment banker might have the incentive to underprice, since by doing so he can increase the probability of being able to sell the whole issue to outside investors (thereby minimizing his underwriting risk) while earn ing a high investment banking spread (OP - BP) on the issue.4 Clearly, if this was the prime reason for underpricing, it would tend to make a case for allowing commercial banks into the under writing business. This argument would be based on the expectation that pro-competitive effects would reduce the average degree of underpricing.5 But this argument would, of implicitly, this argument presumes that investment bankers are risk-averse. This is reasonable, given the pri vate nature of many companies, their limited capital bases, and the potential for a large loss if they take a "big hit" (loss) on an underwriting. For example, many U.S. investment bankers suffered significant losses in underwriting an issue of British Petroleum shares at the time of the October 1987 stock market crash. 5A different monopoly-based argument, advanced in Baron (1982), is that investment bankers possess monopoly power through their private access to information about the likely size of the demand for a new issue. Since issuers are viewed as being relatively uninformed about the nature of this demand, they can easily be exploited by the investment banker. Indeed, since the issuer has no way of knowing ex ante the size of investor demand, the underwriter has an incentive to save resources on distribution and search ("shirking") by simply underpricing enough to ensure that the whole issue is sold. In this context, the presence of potential competitors, such as commercial banks, and the importance of maintaining a reputation might be viewed as potential controls on the investment bankers' temptation to shirk. This presumes, however, that commercial banks, if they entered into underwriting, have the same abilities to "place" (sell to investors) a new issue as investment bankers do. In reality, it might take commercial bankers a number of years to build up the same placement powers. FEDERAL RESERVE BANK OF PHILADELPHIA Why Are So Many New Stock Issues Underpriced? course, be tempered by the need to maintain safety and soundness of the banking system, which could be lessened if the spread (P - OP) is small enough to risk inability to sell the entire issue.6 Due-Diligence Insurance. A second reason given for why underwriters underprice IPOs is the fear of potential legal problems stemming from overpriced issues. Underwriters, along with company directors, are required to exer cise "due diligence" in ensuring the accuracy of the information contained in the prospectus they offer to investors.7 Since passage of the Securities Acts of 1933 and 1934, both under writers and directors may be held legally re sponsible under SEC regulations for the accu racy of this information. Investors who end up holding heavily over priced issues may well have an incentive to sue the underwriter and/or the company directors for publishing misleading or incomplete infor mation in the prospectus. The investors could contend they were misled into believing this was a "good" issue rather than a "bad" one. To avoid any negative legal effects, as well as adverse publicity and damage to reputation, a risk-averse underwriter may try to keep inves tors happy by persistently underpricing IPOs. Hence, some researchers believe that the legal penalties for due-diligence failures are what have created incentives for investment bankers to underprice. The Problem of the "Winner's Curse." The academic literature has paid a great deal of attention to a theory first advanced by Rock (1986) and extended by Beatty and Ritter (1986) and McStay (1987), among others. This theory considers underpricing as a competitive out 6 Since P is not known with certainty, a small spread (P - OP) risks occasional negative spreads, in which case the underwriting firm suffers a loss. 7 See, for example, Tinic (1988). Anthony Saunders come in an IPO market in which some investors are viewed as informed while a larger group is viewed as uninformed. As a result, underpricing is directly related to the degree of infor mation imperfection—or, more specifically, information asymmetry—in the capital market and to the costs of collecting information. Both this theory and the one that follows view un derpricing as a way of resolving the problem of costly information collection. In Rock's model, there are two types of IPOs: good issues and bad issues. Informed investors, defined as those who expend re sources collecting information on IPOs, will bid only for those issues that are good. (This search effort is assumed to allow the informed investor to assess exactly the true value of the IPO.) Those investors who are uninformed, however, will not engage in expensive search, but rather will bid randomly across all issues, good and bad. It is further assumed that informed investors are never sufficiently large as a group to be able to purchase a whole issue. First, consider a good issue. In this case, both informed and uninformed investors will bid for the issue (the uninformed in a random manner). Because both groups bid for the issue, it is likely to be oversubscribed, so that any single individual bidder (informed or unin formed) will get fewer shares than he bid for. Thus, for good issues, uninformed investors get only partial allotments. Next, consider bad issues. In this case, informed investors will not bid at all. The only bidders will be the uninformed. Moreover, owing to the absence of competing informed bidders, any individual bidder will more likely achieve his full allotment (or a higher probabil ity of an allotment). That is, the uninformed bidder suffers from the problem of the "win ner's curse": he achieves a large allotment for bad IPOs and a small allotment for good IPOs. Rock's argument is that, because of the win ner's curse, IPOs have to be underpriced on average so as to produce an expected return for 7 BUSINESS REVIEW the uninformed investor that is high enough to attract investment in IPOs regardless of whether the issue is good or bad.8 That is, underpricing is a phenomenon perfectly consistent with competitive market conditions in a world of imperfect information flows. Thus, monopoly power is rejected as an argument explaining underpricing. Underpricing as a Dynamic Strategy. In the most recent literature, underpricing is seen as a dynamic strategy employed by issuing firms to overcome the asymmetry of informa tion between issuing firms and outside inves tors.9 Implicitly, underpricing is viewed as a cost to be borne by the issuing firm's insiders to persuade investors to collect (or aggregate) in formation about the firm and in that way estab lish its true value in the secondary market. Moreover, the better the firm (a "good" issue), the more it will be underpriced relative to the bad issue. Specifically, a good firm will underprice its issue to attract outside investors.10 Investors (such as analysts) collect information about the firm and, in the secondary market, establish its true value above its offer price. The owners of the firm benefit from this strategy because once the true (higher) market value is established, the owners have an incentive to "cash in" by coming out with new (further) secondary of ferings at the higher market price. Thus, the cost or losses of underpricing the IPO are offset MARCH/APRIL 1990 by the benefits from cashing in on the secon dary offering.11 By comparison, a bad firm—one that knows it is a bad firm—will have the opposite incen tives. In particular, the firm may seek to price the IPO as high as possible, since it knows that once investors collect information and discover that it is a "bad" firm, its stock's price will fall on the secondary market.12* As in the Rock model, these types of dy namic-strategy models view underpricing as a phenomenon that is consistent with competi tion in a world of imperfect information among issuing firms and investors. The difference is that, here, IPO underpricing is viewed as a cost to be borne by good firms, which is offset by the revenue benefits from making a secondary ("seasoned") offering later on at a higher price. IMPLICATIONS FOR BANK REGULATION What do these models imply for bank regu lation and, in particular, the Glass-Steagall Act? If underpricing is indeed due to information imperfections in the capital market—especially between firms and investors—it is difficult to see how commercial banks' entry into under writing will have much effect, unless these banks somehow collect more information and alleviate the degree of information imperfec tion in the market. Since the modern theory of banking views banks as major collectors and users of information, increased production of information about small firms may indeed be a benefit from repealing Glass-Steagall. However, a better test of whether Glass- 8 Technically, the conditional expected return for the un informed investor, across both good and bad issues, must be at least as great as the risk-free rate. 9 See, for example, Chemmanur (1989) and Welch (1988). 10 In these models, the investment banker plays a largely passive function, operating as an agent on behalf of the principal (the firm). The failure of the investment banker to take a more active role may be seen as a weakness of these information-based models. 8 11 Welch (1988) offers preliminary evidence that these issues that are more underpriced tend to follow up more quickly with a secondary (seasoned) offering. 12 This is not to imply that the bad firms necessarily over price. However, the theory has the aggregate implication that the greater the porportion of good to bad issues in the market, the greater the degree of underpricing on average. FEDERAL RESERVE BANK OF PHILADELPHIA Why Are So Many New Stock Issues Underpriced? Steagall has undesirable costs is whether it confers monopoly power on existing invest ment banks that is reflected in the degree of underpricing. That is, what, if any, is the empirical evidence linking underpricing to the monopoly power of investment banks? One implication of the monopoly-power hy pothesis13is that an underwriter, because of his expertise and more precise knowledge of the issuing firm's true value, can save effort (shirk) by ensuring maximum sales through underpricing while still earning a high underwriting spread (OP - BP). However, even in a world of asymmetric information, presumably firms would learn that they are being exploited and, if competition exists, would switch to other underwriters. In contrast, monopoly power would imply that issuing firms would fare as well with one investment bank as with another and that underwriters could ignore all prob lems or considerations related to maintaining a reputation. Beatty and Ritter (1986) have sought to test this reputation-monopoly power effect. That is, do investment bankers who heavily under price in one period lose business from issuing firms in the next? Beatty and Ritter's results tended to confirm that the more an investment banker underpriced in one period, the greater his loss of business in the next—a result sug gesting that monopoly power is temporary at best. A second implication of the monopoly-power hypothesis is that the investment banker—to avoid risk—will have a greater incentive to underprice relatively risky issues so as to en sure maximum sales. For example, it can be argued that the more uncertain are firms' uses of the proceeds of the issue (for example, to pay Anthony Saunders off existing debt, to develop new projects, and so on), the riskier the issue. Or, alternatively, the more variable the after-market returns on an issue— measured by the standard deviation of returns over a period subsequent to listing on the stock exchange— the riskier the issue. Thus, we would expect underpricing to in crease as the number of potential uses of pro ceeds, and the volatility of its (expected) price in the after-market, grows. Beatty and Ritter (1986) found a positive relationship between number of uses of pro ceeds and underpricing; Ritter (1984) and Miller and Reilly (1987) found a positive relationship between the standard deviation of after-mar ket returns and the degree of underpricing. Both these results are consistent with the monopoly-power hypothesis; however, it must be noted that both findings are also consistent with the competitive-market, informationimperfection "winner's curse" theory of Rock (1986).14 A third potential implication of the monop oly-power model is that the degree of underpricing should have been less prior to passage of Glass-Steagall— that is, the pre-1933 average degree of underpricing should have been less than the post-1933 average degree. In a recent study, Tinic (1988) tested the degree of underpricing in the period 1923-30 and compared it with the period 1966-71. He found that underpricing was higher in the 1966-71 period. While Tinic interpreted these results as consistent with the due-diligence-insurance hypothesis— that is, the passage of the Securities Act of 1934, which forced investment banks to underprice to avoid potential lawsuits— they are also con sistent with the monopoly-power hypothesis. That is, in a period preceding Glass-Steagall (when commercial banks had greater power to 13 See Baron (1982), who developed a theory of invest 14 That is, the greater the risk or uncertainty about the ment banker monopoly power based on the inability of issue, the greater the cost of becoming informed and thus the issuers to accurately monitor the investment bankers' effort greater the degree of underpricing required in equilibrium. in placing new shares with investors. 9 BUSINESS REVIEW MARCH/APRIL 1990 underwrite corporate securities),15 the degree than the median or mean underpricing found of underpricing was less than in a period fol in the majority of studies listed in the table lowing the Glass-Steagall separation of pow below and that monopoly power may offer a partial explanation for underpricing. ers. A fourth implication of the monopoly-power Nevertheless, the results favoring monop hypothesis is that IPOs of investment banks oly power as the major determinant of new(for example, Morgan Stanley going public) issues underpricing appear somewhat weak. should not be underpriced, since the invest Indeed, the evidence is largely consistent with ment bank brings its "own firm" public. Look the existence of competitive markets in which ing at 37 IPOs of investment banks that went investors have incomplete or imperfect inforpublic in the 1970-84 period and partici pated in the distri Initial Returns, According to Various Studies bution of their own issues, Muscarella Sample Sample Initial Returns and V etsuypens Period Size 1 Week 1 Mo. Study (1987) find an aver age degree of underReilly/Hatfield (1969) 53 9.9% 8.7% 1963-65 pricing of 8 percent McDonald/Fisher (1972) 1969-70 142 28.5% 34.6% on the first day of — 1965-69 41.7% Logue (1973) 250 trading. At first sight — 62 9.9% Reilly (1973) 1966 this tends to contra 17.1% 19.1% Neuberger/Hammond (1974) 1965-69 816 dict the monopoly power hypothesis as — 1960-71 Ibbotson (1975) 128 11.4% the sole reason for Ibbotson/Jaffe (1975) 1960-70 2650 16.8% — underpricing; how 10.9% 1972-75 486 11.6% Reilly (1978) ever, it could be 102 5.9% Block/Stanley (1980) 1974-78 3.3% argued that 8 percent Neuberger/LaChapelle (1983) 1975-80 118 27.7% 33.6% underpricing is less 15This was particu larly true in 1927-33, when commercial banks had the same powers as investment banks. Since technology and the struc ture of the financial serv ices industry are continu ously changing, a more valid test might have been to compare underpricing in the period im mediately following pas sage of the Glass-Steagall Act. 10 Ibbotson (1982) 1971-81 N /A Ritter (1984) 1960-82 5162 — 18.8% 2.9% — 1977-82 1028 26.5% — 1980-81 325 48.4% — Giddy (1985) 1976-83 604 10.2% John/Saunders (1986) 1976-82 78 — — 8.5% Beatty/Ritter (1986) 1981-82 545 14.1% — Chalk/Peavy (1986) 1974-82 440 13.8% — Ritter (1987) 1977-82 Firm commitment 664 14.8% — Best efforts 364 47.8% — 9.9% — — 7.6% Miller/Reilly (1987) 1982-83 510 Muscarella/Vetsuypens (1987) 1983-87 1184 FEDERAL RESERVE BANK OF PHILADELPHIA Why Are So Many New Stock Issues Underpriced? Anthony Saunders mation about new firms. While new issues did appear to be less underpriced before GlassSteagall (consistent with the monopoly-power hypothesis), evidence suggests that those in vestment banks that excessively underprice today lose future business from prospective issuing firms and that investment banks' own IPOs are also underpriced on average (although less so than those of other firms). The gains from allowing commercial banks to compete directly with investment banks for corporate equity underwritings may come less from cre ating more potential competition than from collecting, producing, and disseminating more information about small firms in the new-issue process. This conclusion suggests that allow ing banks into investment banking activities may indeed bring about price changes that benefit the public; however, those changes may be smaller and occur for different reasons than once thought. REFERENCES Baron, D.P. "A Model of the Demand for Investment Banking and Advising and Distribution Serv ices for New Issues," Journal of Finance (1982) pp. 955-77. Beatty, R., and J. Ritter. "Investment Banking, Reputation, and the Underpricing of Initial Public Offerings," Journal of Financial Economics (1986) pp. 213-32. Block, S., and M. Stanley. "The Financial Characteristics and Price Movement Patterns of Companies Approaching the Unseasoned Securities Market in the Late 1970s," Financial Management (1980) pp. 30-36. Chalk, A.J., and J.W. Peavy. "Understanding the Pricing of Initial Public Offerings," Southern Methodist University Working Paper 86-72 (1986). Chemmanur, T.J. "The Pricing of Initial Public Offerings: A Dynamic Model With Information Production," mimeo, New York University (1989). Giddy, I. "Is Equity Underwriting Risky for Commercial Bank Affiliates?" in I. Walter, ed., Deregulat ing Wall Street (New York: John Wiley, 1985). Ibbotson, R.G. "Price Performance of Common Stock New Issues," Journal of Financial Economics 3 (1975) pp. 235-72. Ibbotson, R.G. "Common Stock New Issues Revisited," Graduate School of Business, University of Chicago, Working Paper 84 (1982), unpublished. Ibbotson, R.G., and J.J. Jaffe. " ’Hot Issue' Markets," Journal of Finance 30 (1975) pp. 1027-42. John, K., and A. Saunders. "The Efficiency of the Market for Initial Public Offerings: U.S. Experience 1976-1983," unpublished (1986). Logue, D.E. "On the Pricing of Unseasoned New Issues, 1965-1969," Journal of Financial and Quanti tative Analysis (1973) pp. 91-103. ll BUSINESS REVIEW MARCH/APRIL 1990 McDonald, J.G., and A.K. Fisher. "New-Issue Stock Price Behavior," Journal of Finance (1972) pp. 97-102. McStay, K.P. The Efficiency of New Issue Markets, Ph.D. thesis, Department of Economics, U.C.L.A. (1987). Miller, R.E., and F.K. Reilly. "An Examination of Mispricing, Returns, and Uncertainty for Initial Public Offerings," Financial Management (1987) pp. 33-38. Muscarella, C.J., and M.R. Vetsuypens. "A Simple Test of Baron's Model of IPO Underpricing," Southern Methodist University Working Paper 87-14 (1987a). Muscarella, C.J., and M.R. Vetsuypens. "Initial Public Offerings and Information Asymmetry," Edwin L. Cox School of Business, Southern Methodist University, unpublished (1987b). Neuberger, B.M., and C.T. Hammond. "A Study of Underwriters' Experience With Unseasoned New Issues," Journal of Financial and Quantitative Analysis (1974) pp. 165-77. Neuberger, B.M., and C.A. LaChapelle. "Unseasoned New Issue Price Performance on Three Tiers: 1975-1980," Financial Management (1983) pp. 23-28. Pugel, T.A., and L.J. White. "An Empirical Analysis of Underwriting Spreads on IPO's," Working Paper 331, Salomon Brothers Center for the Study of Financial Institutions, Graduate School of Business Administration, New York University (September 1984). Reilly, R.K., and K. Hatfield. "Investor Experience With New Stock Issues," Financial Analysts Journal (September/October 1969) pp. 73-80. Reilly, R.K. "Further Evidence on Short-Run Results for New Issue Investors," Journal of Financial and Quantitative Analysis (1973) pp. 83-90. Reilly, R.K. "New Issues Revisited," Financial Management (1978) pp. 28-42. Ritter, J. "The 'Hot Issue' Market of 1980," Journal of Business 57 (1984) pp. 215-40. Ritter, J. "The Costs of Going Public," University of Michigan Working Paper 487 (1987a). Ritter, J. "A Theory of Investment Banking Contract Choice," University of Michigan Working Paper 488 (1987b). Rock, K. "Why New Issues Are Underpriced," Journal of Financial Economics 15 (1986) pp. 187-212. Tinic, S. M. "Anatomy of Initial Public Offers of Common Stock," Journal of Finance 43 (1988) pp. 789-822. Welch, I. "Seasoned Offering, Imitation Costs and the Underpricing of Initial Public Offerings," University of Chicago Working Paper (1988). 12 FRASER Digitized for FEDERAL RESERVE BANK OF PHILADELPHIA Is There a Natural Rate of Unemployment? William W. Lang * The idea that a nation's unemployment rate gravitates toward some "natural" rate has been a mainstream theory in macroeconomics for the past 20 years. According to this theory, the natural rate of unemployment is determined by factors related to the economy's supply side, such as labor force demographics. Ac tions that influence aggregate demand, such as monetary and fiscal policies, can affect how "■William W. Lang is an Assistant Professor of Economics at Rutgers University, New Brunswick, NJ. He wrote this article while he was a Visiting Scholar in the Research Department of the Federal Reserve Bank of Philadelphia. much unemployment varies over the business cycle, but they cannot affect its average level. The unemployment situation in Europe has forced economists to reevaluate the natural rate theory. During the early 1980s, recessions in Europe and the United States boosted unem ployment rates to their highest levels since the Great Depression. Since then, unemployment has returned to more normal levels in the United States. But in Europe, unemployment remains high even now. Citing the European experience, some econo mists are advocating that the natural rate the ory be replaced with a theory of "hysteresis," a theory that explains how aggregate-demand 13 BUSINESS REVIEW policies can permanently raise the unemploy ment rate. Although the debate is still in its early stages, at issue is the long-run impact of demand-management policies on unemploy ment. MARCH/APRIL 1990 rienced workers find themselves unemployed when their skills are no longer in demand because of declining demand for the goods they once produced or because of changes in technology. For example, in the United States the demand for steel workers has been de pressed since the mid-1970s. Now these struc turally unemployed workers must either relo cate or develop new skills in order to find jobs. According to the natural rate theory, the average level of both structural and frictional unemployment is relatively unaffected by monetary or fiscal policies. Over the long run, THE NATURAL RATE THEORY Regardless of how healthy the economy is, some unemployment is inevitable. Some people quit their jobs, some workers are fired, and some industries reduce employment levels while others increase them. The unemployment that these shifts create constitutes the nation's natu ral rate of unemployment.1 Frictional and Structural Un employment. One component Unemployment Rates of the natural rate is "frictional" in Europe and the U.S. unemployment, represented by unemployed workers who are over the Past Two Decades temporarily between jobs or who have just come into the la After peaking in 1982, U.S. unemployment has returned to bor force. A worker who quits more normal levels of between 5 percent and 6 percent— levels thought to represent the "natural" rate of unemployment. But his job to find work in another unemployment in Europe kept rising after 1982 and has re trade or another industry would mained high. It's easy to see why many economists have ques be considered frictionally un tioned the natural rate theory, at least for the European econo employed. mies. The other component of the natural rate is "structural" un employment. This occurs when workers do not have the neces sary skills to meet the current demands of employers. Often, young workers lack sufficient education or training to find work. Sometimes, even expe 1Edmund Phelps, "Phillips Curves, Expectations of Inflation and Optimal Unemployment Over Time," Journal of Political Economy (August 1967), and Milton Friedman, "The Role of Mone tary Policy," American Economic Review (March 1968), are generally credited with developing the natural rate theory of unemployment. 14 FEDERAL RESERVE BANK OF PHILADELPHIA Is There a Natural Rate of Unemployment? frictional and structural unemployment is de termined by supply-side factors: the demo graphic composition of the labor force, shifts in employment between industries and regions, minimum-wage laws, and government bene fits to the unemployed. The demographic composition of the labor force can affect the natural rate of unemploy ment significantly. For example, workers under the age of 25 have higher average rates of unemployment than older workers. This is because young workers change jobs relatively frequently in their search for appropriate ca reer employment. In other words, young workers have higher rates of frictional unem ployment because they either quit or are fired more frequently than older workers. Rapid shifts in employment across indus tries also tend to increase the levels of struc tural and frictional unemployment. When workers must shift from one industry to an other, they usually experience a period of unemployment while searching for new jobs. Moreover, workers in declining industries may not have the appropriate skills for the indus tries that do have job openings. All of this would lead to higher levels of structural unem ployment. Increases in minimum-wage benefits and in government benefits to the unemployed tend to increase the level of structural and frictional unemployment. A higher minimum wage will increase structural unemployment, since em ployers are less inclined to hire poorly edu cated workers with little work experience if they must be paid a higher wage. Similarly, an increase in government benefits to the unem ployed will increase frictional unemployment, as these higher benefits tend to make unem ployed workers less willing to accept lowerpaying jobs. Cyclical Unemployment. According to the natural rate theory, normal rates of structural and frictional unemployment are invariant to demand-management policies, but the cyclical William W. Lang component of unemployment is not.2 For example, a contractionary monetary policy lowers the demand for goods and serv ices, which tends to reduce inflation. But wage increases do not slow commensurately. Work ers' wages are often set a year or so in advance, many by contract. And rarely are wages in dexed completely to the inflation rate.3 So firms, faced with declining demand for their product and inflexible labor costs, lay off work ers and cut back on output. Thus, the tighter monetary policy reduces inflation but raises unemployment. According to the natural rate theory, this trade-off between inflation and unemployment is short-lived. The economy eventually adjusts to the lower inflation rate. Workers and firms write new wage contracts based on the lower inflation rate, and real wages once again reflect the fundamental supply and demand condi tions. Producers find it profitable to rehire workers and raise output. In the end, the con tractionary monetary policy permanently low ers the inflation rate, but unemployment re turns to its natural rate. THE NATURAL RATE THEORY EXPLAINS THE U.S. EXPERIENCE— BUT NOT EUROPE'S The natural rate theory associates an eco nomic downturn with declining inflation and 2A s already discussed, a government's social welfare policy may have an impact on the levels of structural and frictional unemployment. 3There are, of course, different versions of the natural rate theory. Modern Keynesians emphasize the role of rigid wages and prices in explaining the short-run effects of monetary policy. The New Classical economists argue that only the unexpected components of monetary policy affect unemployment. However, the crucial issue for our discus sion is that both versions of the natural rate theory have argued that monetary policy, over the long run, has no effect on the average unemployment rate. 15 BUSINESS REVIEW MARCH/APRIL 1990 How Different Are the Inflation-Unemployment Experiences for the U.S. and Europe? These graphs show the unemployment rate (horizontal axis) and the inflation rate (vertical axis) for the United States, France, West Germany, and the United Kingdom over the 1979-88 period. Note the U.S. economy's proclivity to return to a "natural rate" of unemployment: U.S. unemployment increased from 1979 to 1982 while inflation had begun to decline in 1980; after inflation had stabilized by 1983, unemployment reversed direction to settle in 1988 at a rate even lower than 10 years before. Inflation also plummeted in the three European economies; however, unemployment there has actually increased in the past decade. 0 _____ I______ I______ I______ I______ I_____ I 4 5 6 7 8 9 10 Unemployment Rate (Percent) FEDERAL RESERVE BANK OF PHILADELPHIA William W. Lang Is There a Natural Rate of Unemployment? high unemployment in the short run. Over the longer run, unemployment returns to the natu ral rate while inflation remains low. The recent paths of inflation and unemploy ment in the U.S. fit the natural rate theory. Between 1979 and 1982, the United States adopted disinflationary money and credit poli cies that drove the unemployment rate to nearly 10 percent. But since 1983, inflation has leveled off and unemployment has fallen even below its 1979 level. While its precise level is debat able, the natural rate of U.S. unemployment seems to be somewhere between 5 percent and 6 percent. The inflation and unemployment levels for several European countries tell a dramatically different story. For example, between 1979 and 1985 the inflation rate in France fell and unemployment rose. But after 1985, when the inflation rate stabilized at about 3 percent, the unemployment rate remained in double digits. While the French example is the most dra matic, the unemployment rates for West Ger many and the United Kingdom seem to have stabilized at significantly higher levels as well. HAS EUROPE'S NATURAL RATE RISEN? In principle, changes in nor mal rates of structural or fric tional unemployment could have boosted Europe's natural rate above the U.S. rate. However, the factors most commonly cited fail to support this view. Demographic Changes. One explanation commonly given for the notion of a higher natural rate in Europe is the increase in the relative shares of women and youth in the labor force.4 The 4The argument for higher average rates of unemployment among women may be losing some of its force, as women structural and frictional unemployment rates for both groups are higher than those for male workers. However, demographic changes alone do not justify a relative rise in many European countries' average unemployment rates. Let's look at the change in the share of young workers in the U.S., French, German, Italian, Dutch, and Swedish labor markets between the 1960s and the 1980s (see Figure 1). Of all these countries, only the U.S. has seen an increase in that share. The European countries have expe rienced declining shares.*5 Women's labor force shares increased for all the countries (see Figure 2). However, there is no strong correlation between those countries showing large gains in this share and those workers are becoming more firmly attached to the labor force. In fact, in the U.S., the female unemployment rate is now roughly equivalent to the male unemployment rate. 5This may seem surprising, since European countries also experienced a baby boom. However, young people in Europe stay in school longer and thus remain out of the labor force longer than their U.S. counterparts. FIGURE 1 Changes in Youths’ Share of Total Labor Market from the 1960s to the 1980s Percent 12-, - 8-1 US France Germany Italy Netherlands Sweden 17 BUSINESS REVIEW MARCH/APRIL 1990 to another increases unemploy ment, because workers in the Changes in Women’s Labor Market declining industries must learn Share from the 1960s to the 1980s new skills and search for jobs in healthier industries.6 Percent On the surface, this explana tion seems to fit with the devel oped nations' rapid shifts in employment from the manufac turing sector to the service sec tor. But several studies have found that sectoral shifts have little impact on overall unem ployment rates.7* To identify national shifts in employment, economists often develop a "mismatch index" for the country. The index captures US France Germany Italy Netherlands Sweden the divergences in employment growth among the industries in countries with large increases in unemploy the economy. When the index number is low, ment. Sweden, which has not experienced employment rates in all industries are growing sustained high unemployment, shows the larg at about the same pace. When the index num est expansion in the percentage of working ber is high, the industries' employment growth women. The U.S. is in the middle in terms of rates are diverging significantly—some may growth in the share of women in the labor force be growing rapidly, some more slowly, and when compared to France, Germany, Italy, and some not at all. Having calculated mismatch indexes for the Netherlands; yet all of those European countries have experienced greater increases several OECD countries, researcher Robert Flanagan found that the indexes for France and in unemployment rates. West Germany were lower in the late 1970s In short, the European countries with large than they had been in the previous 15 years and increases in average unemployment have not seen relatively big increases in the shares of women and young people in their labor forces. This suggests that demographic shifts are not inducing a higher natural rate of unemploy 6Probably the most influential paper on this subject is by ment for these countries. David Lilien, "Sectoral Shifts and Cyclical Unemploy Job Shifts Across Industries. Another popu ment," Journal of Political Economy 90 (August 1982) pp. 777lar explanation for why Europe's natural rate 93. may have risen is that the secular movement of 7See, for example, Richard Jackman, Richard Layard, employment away from manufacturing and and Christopher Pissarides, "On Vacancies," Discussion toward services has increased both structural Paper 165, London School of Economics, Centre for Labour and cyclical unemployment. Even if the aver Economics, 1985; Orley Ashenfelter and David Card," Why Have Unemployment Rates in Canada and the United age number of jobs to be filled remains un States Diverged?" Economica 53 (Supplement 1986) pp. 171changed, shifting employment from one sector 96. Digitized for 18 FRASER FIGURE 2 FEDERAL RESERVE BANK OF PHILADELPHIA Is There a Natural Rate of Unemployment? that they remained low in the 1980s. And while the index value for the United Kingdom rose in the early 1980s, the rate of increase only matched that of the U.S., where unemploy ment has fallen back to its earlier levels.8 (See Mismatch Indexes.) So, at least by this measure, a shifting industrial mix does not seem to have raised natural rates of unemployment for France, West Germany, and the United Kingdom re cently. Flanagan examined other types of mismatch indexes, including measures of shifts in labor market conditions across regions. He con cluded that any mismatch effect on the relative rise in European unemployment has been small. Minimum-Wage Laws and Government Pro grams. Government programs and minimumwage laws, the last factors cited as possibly having an effect on the natural rate of unem ployment, could in principle have generated increases in Europe's natural rate. There is little evidence, however, that Europe's minimum-wage laws and government benefits to the unemployed have been more generous over the last decade than before. If anything, gov ernment programs have tended to be more stringent than in the past. Government programs may be contributing to the increase in European unemployment indirectly. When there are employment de clines in high-wage industries, workers may be unwilling to accept low-wage jobs if they are receiving substantial unemployment benefits from the government. But it would be difficult to argue that government benefit programs are the sole explanation for high European unem ployment. 8Robert J. Flanagan, “Labor Market Behavior and Euro pean Economic Growth," in Barriers to European Growth, Robert Z. Lawrence and Charles L. Schultze, eds. (Washing ton, D.C.: Brookings Institution, 1987) pp. 175-211. William W. Lang Mismatch Indexes Divergences in Employment Growth Among Industries, 1960 -1983 Years 1960-64 1965-69 1970-74 1976-79 1980-83 France 2.3 2.8 2.8 2.0 1.7 West Germany U.K. U.S. 2.6 3.2 3.2 1.9 1.8 1.9 2.2 2.5 1.4 3.4 2.4 2.1 2.3 1.5 2.9 Based on data in Robert J. Flanagan, "Labor Market Behavior and European Growth," in Barriers to European Growth, Robert Z. Law rence and Charles L. Schultze, eds. (Brookings Institution: Washington, D.C., 1987) p. 181. Data for 1975 were not available. THEORIES OF HYSTERESIS: THE IMPACT OF AGGREGATE-DEMAND POLICIES The prolonged slump in Europe has helped revive the notion that aggregate-demand pol icy can have long-run impacts on the level of unemployment.9 The term "hysteresis" has been used to describe theories in which tempo rary shifts in aggregate demand cause perma nent or long-term changes in unemployment. If there is a natural rate of unemployment, 9Unemployment, Hysteresis and the Natural Rate Hypothe sis, Rod Cross, ed. (Basil Blackwell Ltd., 1988), presents various papers on this subject by European and American authors. In "On the History of Hysteresis," Rod Cross and Andrew Allan note that the term "hysteresis" comes from the Greek word meaning "to come after" or "to be behind." It was originally used by physical scientists to describe the tendency for a previous state or condition to persist. Econo mists apply the term to theories that attempt to explain why high unemployment in one period tends to produce high unemployment in subsequent periods. 19 BUSINESS REVIEW then a recession causes only temporary changes in unemployment. The unemployment rate returns to the natural rate during the subse quent economic expansion. If there is hyster esis in the unemployment rate, the unemploy ment rate remains permanently higher. That is, there is no inherent tendency for the unem ployment rate to fall back to its pre-recession level.10 Wage Rigidities. Almost all theories of hysteresis in unemployment have in common the notion that real wages are not fully flexible, even in the long term. For one reason or another, real wages remain high even when there are large numbers of unemployed work ers willing to work for less. Recent discussions of hysteresis have fo cused on microeconomic rationales for wage rigidity. In particular, theorists are exploring the idea that employed workers have the power to prevent wage cuts and thus introduce the rigidities that cause hysteresis in unemploy ment. Insider/Outsider Models. In insider/out sider models, employed workers, called "in siders," are able to maintain wages at high levels even though unemployed workers, or "outsiders," are willing to work for lower wages. Insiders can prevent firms from hiring the lowwage workers by making it more costly for firms to fire employees and hire others in their place. To some extent, every firm that hires new workers incurs some cost in training them. Insiders can raise the costs by refusing to par ticipate fully in the training process. And they can punish firms that hire outsiders at low wages in other ways. They can take some overt action to disrupt production, such as staging a strike or a slowdown. Or they may simply put 10Of course, an expansionary economic policy or other positive economic events may push the unemployment rate below its pre-recession level. MARCH/APRIL 1990 less effort into their jobs.11 Forming a labor union can enhance insiders' power to act col lectively, but insiders can punish the firm even without a union. The key insight of the insider/outsider models is that once workers become unemployed, they lose their status as insiders. The now-smaller group of insiders is unwilling to reduce wages in order to get the unemployed rehired, be cause these former employees no longer exer cise any influence in the group. In other words, the more exclusive the group, the less willing the group will be to make wage concessions to increase employment. The Permanent Impact of Aggregate De mand. To explain permanent shifts in the unemployment rate, theories of hysteresis add to the insider/outsider model the notion that aggregate-demand swings can cause persis tent productivity shifts. Suppose a monetary contraction slows the economy and induces firms to lay off workers. According to the insider/outsider model, the remaining insiders will keep real wages from falling, despite the slack in the labor market. But if the economic contraction somehow re duces labor's productivity, then firms will not rehire laid-off workers unless real wages de cline. So the combination of rigid wages and lower productivity keeps unemployment from returning to its old level. Now the question is, how does the economic contraction perma nently lower productivity? n For a fuller treatment of the insider/outsider model, see Assar Lindback and Dennis Snower, "Wage Setting, Un employment and Insider-Outsider Relations," American Economic Review 72 (1986). Note that the ability of insiders to punish firms for hiring low-wage workers would not affect new entrants into an industry. Such insider power will play a significant role only when there are fixed costs or other barriers to start-up firms in an industry. It is worth noting that much of the increase in U.S. employment over the past decade is due to small firms, many of them start-up firms. This may help explain why insiders in the U.S. have been less effective in exerting pressure to maintain high real wages. FEDERAL RESERVE BANK OF PHILADELPHIA Is There a Natural Rate of Unemployment? Theorists offer two explanations. First, a contractionary monetary policy raises interest rates and lowers spending on capital goods. The reduction in capital formation, in turn, lowers workers' productivity. Without a commensurate fall in real wages, firms have less incentive to hire workers—and so unem ployment is permanently higher. What is the evidence for this source of hys teresis? While member nations of the Euro pean Economic Community have seen sub stantial increases in their unemployment rates, the ratio of capital to employed worker has remained roughly constant. This has led some to argue that the existing capital stock is inade quate to employ the current available labor force in the EEC countries. The second possible explanation for hyster esis focuses on the long-term impact of unem ployment itself. When an economic contrac tion throws people out of work, long layoffs may erode their job skills. Without a decline in real wages, these less skilled workers will find firms unwilling to rehire them. Thus, what could have been a temporary increase in unem ployment is perpetuated by the wage rigidity. The argument that prolonged unemploy ment will erode job skills is difficult to quan tify. Direct measures of labor productivity reflect the productivity of workers who are employed, not those who are unemployed. One piece of supporting evidence for this hypothesis is that a large part of the increase in unemployment is due to an increase in the number of long-term unemployed. In short, theories of hysteresis propose that Europe's high unemployment is due to wage rigidity, insufficient capital formation, and deteriorating job skills. HYSTERESIS LEAVES SOME QUESTIONS UNANSWERED While theories of hysteresis seem consistent with some aspects of the European experience, some difficult issues must still be addressed William W. Lang before these theories gain wide acceptance. First, the data indicate that the persistence of unemployment has increased over the past 20 years both in Europe and in the United States. Since the current theories of hysteresis rely on various forms of wage rigidity, we would expect those rigidities to have increased as well. But there is little evidence that union or insider power has increased over this period. In fact, labor union power has generally waned over the past two decades in Europe and the United States. Perhaps the more important question is why the natural rate theory seems to fit the United States but not Europe. The microeconomics of labor markets in the U.S. show some important differences compared to European labor mar kets. For example, the U.S. has fewer union members as a percentage of the labor force. In addition, social welfare programs in the United States are, on the whole, less generous than in Europe. Both of these factors tend to reduce real wage rigidity in the United States. So perhaps the labor market in the U.S. more closely approximates the type of labor market envisioned by the natural rate theory. Alternatively, hysteresis may characterize labor markets in both the U.S. and Europe, and their experiences may differ only because of different macroeconomic policies. According to this interpretation, after the 1982 recession the United States decided to reduce unemploy ment at the risk of higher inflation by engaging in a more stimulative macroeconomic policy than Europe. The U.S. has yet to experience a sharp accel eration in inflation. Perhaps this is because the recessions of 1980 and 1982 have given the Federal Reserve credibility as an inflationfighter—and this is keeping the lid on inflation expectations. Oil prices have helped as well. Their sharp increases of the 1970s were largely reversed in the 1980s. Separating the contributions of macroeco nomic demand-side policy from microeconomic 21 BUSINESS REVIEW supply-side conditions is crucial to U.S. poli cymakers. If it is the microeconomics of the labor market that differentiate the U.S. from Europe— that is, if the U.S. has a natural rate of unemployment but Europe does not—then U.S. policymakers face no trade-off between infla tion and unemployment in the long run. If it is macroeconomics that separate the two—in other words, if both the U.S. and Europe are subject to hysteresis— then U.S. inflation policies have a lasting impact on unemployment. CONCLUSION Stubbornly high unemployment rates in Europe are beginning to undermine econo mists' confidence in the natural rate theory. The theory says that only supply-side factors, such as demographics and technology, have MARCH/APRIL 1990 any persistent impact on a nation's unemploy ment rate. There is little evidence, however, that ad verse supply-side shifts have hit Europe in recent years. Now some economists are break ing away from the natural rate idea and are exploring the possibility that aggregate de mand shifts—including changes in monetary and fiscal policy—can have persistent effects on the level of unemployment. According to these theories of hysteresis, Europe's high unemployment is the legacy of policymakers' anti-inflation programs of the early 1980s. If these theories are correct, then policymakers' decisions have much more of a long-run impact on the unemployment rate than economists had realized up until now. FEDERAL RESERVE BANK OF PHILADELPHIA BUSINESS REVIEW Ten Independence Mall, Philadelphia, PA 19106-1574