The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Business Review Federal Reserve Bank of Philadelphia November • December 1996 For Better and For Worse: Three Lending Relationships Mitchell Berlin The Suburban blousing Market: The Effects of City and Suburban Job Growth Richard Voith ISSN 0007-7011 Business Review The BUSINESS REVIEW is published by the Department of Research six times a year. It is edited by Sarah Burke. Artwork is designed and produced by Dianne Hallowell 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 obtained to reprint portions o f articles or whole articles. Permission to photocopy is unrestricted. Please send subscription orders, back orders, changes o f address, and requests to reprint to Publications, Federal Reserve Bank o f Philadelphia, Department o f Research and Statistics, Ten Independence Mall, Philadelphia, PA 19106-1574, or telephone (215) 574-6428. Please direct editorial communications to the same address, or telephone (215) 574-3805. This publication and others are also available on the Internet using the World Wide Web at 'http://www.phil.frb.org'. Digitized for 2 FRASER NOVEMBER/DECEMBER 1996 FOR BETTER AND FOR WORSE: THREE LENDING RELATIONSHIPS Mitchell Berlin Are close, long-term relationships be tween borrowers and lenders feasible in an increasingly competitive financial mar ketplace? How do relationships that have developed betw een banks and firm s change when firms gain access to alterna tive funding sources, especially public se curities markets? Can firms gain the best of both worlds by a judicious mixture of bank and public borrowing? Using three firms as examples, Mitchell Berlin sizes up the pros and cons of relationship lending. THE SUBURBAN HOUSING MARKET: THE EFFECTS OF CITY AND SUBURBAN JOB GROWTH Richard Voith How does the location of new jobs in a metropolitan area affect the suburban housing market? Does it matter whether job growth occurs in the city or in the sub urbs? And who, if anyone, benefits from job growth? Dick Voith takes a look at housing prices and construction rates in some Philadelphia suburbs to determine the impact of employment growth on the value of real estate assets. FEDERAL RESERVE BANK OF PHILADELPHIA For Better and For Worse: Three Lending Relationships Mitchell Berlin* W w w hen bankers speak of building a rela tionship with a business customer these days, they usually mean selling the customer a whole range of financial products such as lock-boxes, letters of credit, and swaps, in addition to loans. When financial economists speak of relationship lending between banks and firms, they usually have a different, more old-fashioned idea in mind. They mean a close relationship between a firm and its banker, in which a single banker has intimate knowledge about the firm's affairs, built up over years of lending. Economists dis tinguish this type of lending from the more *Mitchell Berlin is a senior economist and research ad visor in the Research Department of the Philadelphia Fed. anonymous arm's-length lending, in which insti tutions and individuals provide funds to firms by purchasing their public securities (stocks and bonds). Over the last 10 years, financial economists have accumulated a significant body of empiri cal knowledge about the advantages and dis advantages of relationship lending. Their em pirical studies have provided insights into some basic questions: Are close, long-term relation ships between borrowers and lenders feasible in an increasingly competitive financial market place? How do relationships that have devel oped between banks and firms change when firms gain access to alternative funding sources, especially public securities markets? Can firms 3 BUSINESS REVIEW gain the best of both worlds by a judicious mix ture of bank and public borrowing? For firms m aking financial decisions or banks gauging their markets, these are clearly important questions. These questions are also important ones for policymakers. In develop ing countries and formerly communist coun tries—where financial systems are being created from scratch— these are precisely the types of questions that policymakers must confront when they weigh the relative merits of a bankoriented financial system, like that of Japan, and a securities-oriented financial system, like that of the United States. Even in the sophisticated and highly competitive financial markets of countries like the United States, public policy affects the types of banking relationships that firms and banks form. For example, recent leg islative proposals to provide subsidies to pro mote a secondary market for small business loans— much like the secondary market for mortgages—may increase smaller firms' access to securities markets and loosen their relation ships with banks. Understanding the econom ics of lending relationships between banks and firms can illuminate such policy debates.1 EXCLUSIVE RELATIONSHIPS CAN EASE CREDIT FOR SMALL FIRMS Long-Term Exclusive Relationships Are Beneficial...Midget Widget is a small Midwest ern firm with sales of $10 million and a simple financial structure.2 M idget has only two sources of external funds. Many of the firm's input suppliers offer trade credit; for example, lrrhis article focuses on the empirical literature on bank lending in the United States. I have not always referenced seminal papers when later papers contain good discussions of the preceding literature. See the article by Sudipto Bhattacharya and Anjan Thakor for an excellent critical review of the theoretical literature, and the one by Leonard Nakamura for a discussion of both the empirical and theo retical literature. 4 NOVEMBER/DECEMBER1996 Midget's supplier of elbow sockets accepts payment 30 days after delivery. Midget also borrows on a continuing basis from Little Bank on the Prairie. Although Midget was started with a prayer and a loan from the ow ner's brother-in-law nearly 14 years ago, the firm has been taking out—and repaying—business loans from Little Bank for 10 years. Over this 10-year period, Midget's borrow ing terms have gotten better and better. At the outset, Midget was still struggling to establish its niche in the widget market and was barely profitable. But Midget has yet to miss a pay ment to its trade creditors or its ow ner's brother-in-law. When the firm first applied to Little Bank for a loan, a loan officer from the bank performed an especially careful analysis of Midget's books, made phone calls to the firm's trade creditors to ask about the firm's repayment history, and visited the widget plant to inspect the firm's inventories. In fact, this visit was only the first of many regular visits to the plant by the loan officer in charge of Midget's account. After some careful discussions by the bank's lending committee, Little Bank decided that Midget was a good credit risk, in part because of its exemplary repayment history, but prima rily because it was a promising business that had strong future prospects. The lending com mittee also decided that rather than burden Midget with very high loan payments at the outset—which might backfire and push the firm into early default—the bank would charge a loan rate of only prime plus 3 percent. This rate was not high enough to cover Little Bank's ini tial costs of investigating the firm plus its own funding costs (mainly the costs of paying de 2The story of Midget Widget is based primarily on three important articles, one by Allen Berger and Gregory Udell and two by Mitchell Petersen and Raghuram Rajan. Midget and all other firms and banks in this article are fictional, as are their stories. FEDERAL RESERVE BANK OF PHILADELPHIA For Better and For 'Worse: Three Lending Relationships positors), but the loan committee reasoned that these costs would be made up over time. Midget was a rapidly growing business, and Little Bank's lending committee agreed that even though the loan was risky, it was likely to be just the first in a series of future, more profit able loans. But the initial nonprice contract terms were very stringent, designed to give Little Bank lots of leeway to intervene to protect its money. The loan was structured as a one-year commit ment— so after only one year the bank could freely reevaluate the firm's creditworthiness— and Midget was required to post its accounts receivable as collateral. In fact, payments by those firms receiving trade credit from Midget were made straight to the bank, rather than to Midget. In addition, the loan agreement in cluded numerous restrictive covenants, includ ing the bank's right to veto asset sales by the firm and strict requirements that Midget limit borrowings from other sources. Now, in the 10th year of this borrowing rela tionship, Midget's loan terms are much more attractive than at the outset. The firm now bor rows at prime plus 1 percent, instead of the prime plus 3 percent that it paid initially. In stead of a one-year loan commitment, Midget now has a three-year commitment. Although the contract still has restrictive covenants, Midget's loan commitment is no longer collat eralized, and the firm now receives all payments directly from its customers. Midget can now borrow both more cheaply and without such intrusive bank controls be cause its default risk has dropped over time. Firms that fail are most likely to fail in their first few years of operation. After 14 years, it's clear that Midget is not a fly-by-night firm with a high risk of default. Also, it's now cheaper for Little Bank to lend to Midget. Over the last 10 years, the bank has developed expertise in understanding Midget's financial needs and problems, so new loan agreements and adjust ments to old ones do not trigger the same in Mitchell Berlin tensive evaluation as they did at the outset. Moreover, keeping close tabs on collateral is time-consuming, so a bank's lending costs fall when it feels secure enough to lend without collateral. More attractive contract terms are not the only benefit of the firm's long-term relationship with the bank. Midget's owner feels fortunate to have received a loan commitment at all this year. The regional economy is weak, and as the regional market goes, so goes the market for widgets. Other area firms that have been in business at least as long as Midget have simply been unable to get a loan on any terms from a bank. But, unlike Midget, many of these firms have not had a long-term, exclusive relation ship with a single bank. Little Bank's knowl edge of the ins and outs of Midget's financial condition—through good times and bad—al lows the bank to see M idget's fundamental strengths despite the local economic problems. Little Bank also feels a commitment to help Midget through difficult times. ...But Relationships Are Harder to Form in Highly Competitive Loan Markets. From Little Bank's viewpoint the relationship has devel oped much as the lending committee had hoped initially. The low initial loan rates (relative to the firm's risk of default) and high costs of monitoring Midget at the outset of the relation ship—which initially yielded low profit mar gins—have been replaced by years of hand somely profitable loans. These continuing prof its are rooted partly in the knowledge and ex pertise that Little Bank has built up over its 10year relationship with Midget. Little Bank's greater experience in lending to Midget gives it an advantage over potential competitors for Midget's business, all of whom would find it expensive to reproduce Little Bank's knowledge in a reasonable amount of time. As its own costs of lending to Midget have fallen, Little Bank has passed on some of the cost reduction to Midget through a lower loan rate (and more relaxed contractual controls) and kept some of 5 BUSINESS REVIEW NOVEMBER/DECEMBER1996 the cost reduction for itself as higher profit. Even as Little Bank makes profits, it is hard for any competitor to offer Midget a better deal.3 Just how much profit Little Bank can keep as its lending costs fall—and how Midget's loan rate evolves over time— depends mainly on the number and behavior of Little Bank's competi tors, including both banks and nonbank lend ers (such as finance companies). When compe tition in the loan mar ket is weak and the bank doesn't have to worry so much about a competitor's stealing its customer, the bank can take the entire fu ture custom er rela tionship into account when making a loan pricing decision in a given period. As in Midget's case, a bank can profitably charge loan rates below (riskadjusted) lending costs at the outset of the rela tionship—to keep the risk of early default by their risky borrowers low—knowing that it will be able to charge rates above lending costs as the relationship continues. In markets where competition is weak, the loan rate charged to a customer typically starts low and falls relatively slowly as lending costs fall over the life of the lending relationship. In more competitive loan markets, each bank will be more concerned about a competitor's stealing its customer at any time, which puts strong pressure on the bank to cover its lending costs period by period. Banks don't have the luxury of taking tempo rary losses in the expectation of charging rela tively high rates in the future. So, in highly com petitive markets, the loan rate charged to a customer usually starts high and falls more swiftly over the life of the lending relationship. In Midget's view, more competition would certainly be welcome, as it would put pressure on Little Bank to lower its loan rate now. While g ratefu l for Little Bank's initial commit ment of funds and re sources, Midget now views itself as an estab lished firm that de serves low rates. But in Little Bank's view, its own current profits are merely compensation for its heavy initial ex penditures on evaluat ing and m onitoring Midget's credit risk and for the relatively low loan rates that it charged Midget when it was only four years old and a relatively high-risk firm. Moreover, Little Bank would argue that it wouldn't have been willing to make such a risky loan in the first place without the expecta tion of high profits in succeeding years. And Midget might have had to wait until it had a longer track record to get outside funding. An important lesson of Midget's story is that greater competition in loan markets can have com plicated and surprising effects. Clearly, competition limits a bank's ability to increase loan rates and profits at borrowers' expense. But it also creates difficulties in build ing long-term relationships. In particular, it may be difficult for banks to make risky loans—for example, when firms are young and desperately in need of credit—unless the bank expects that it will ultimately profit over the life of the lend ing relationship. Both the competitive advan Banks in highly competitive loan markets don't have the luxury of taking temporary losses in the expectation of charging relatively high rates in the future. Econom ists would say that Midget is locked into its re lationship with Little Bank, because the bank has an infor mation advantage over its competitors. Of course, other banks might learn something about Midget's creditwor thiness based on Little Bank's willingness to make a loan. But as long as Little Bank's credit-granting decision does not com pletely reveal all relevant inform ation about Midget, Midget will be locked in. FEDERAL RESERVE BANK OF PHILADELPHIA For Better and For Worse: Three Lending Relationships tage held by an incumbent bank, because of its prior relationship with the firm, and relatively noncom petitive loan m arkets increase the bank's profits over the life of a lending relation ship. 4 RELATIONSHIPS BECOME LESS EXCLUSIVE AS FIRMS BECOME LARGER Exclusive Relationships Create Tensions. Middle Marketing (popularly known as 2M) is a closely held firm with $50 million in sales, and it has been borrowing from Regional Bank for 15 years. Regional has been 2M's sole banker and, other than trade credit from its suppliers, 2M's only source of outside funds. While the relationship with Regional has been mutually beneficial, 2M is not completely satisfied. In fact, the firm's Treasurer has become increasingly dissatisfied as he has fielded phone call after phone call from Regional's competitors, who are also seeking to expand their presence in the middle market, and from investment bankers who are trying to convince 2M to go public.5 Although 2M no longer has to post collat eral on its loans, its three-year loan agreement still has extensive covenants and contractual controls that the firm finds increasingly intru- 4This should not be interpreted as an argument that greater competition is a bad thing and should be discour aged, but only that there are both benefits and costs. In addition, monopoly profits are not the only way that a bank can receive compensation for its initial commitment of re sources to a firm. For example, some economists have ar gued that holding equity stakes in firms could serve a simi lar function for a bank, even in highly competitive loan markets. This would require changes in laws that separate banking and commerce, which severely restrict bank eq uity positions in firms that are not in financial distress. Such legal changes might have complicated and far-reaching effects. For example, see the article by Loretta Mester. 5The middle market is a fairly nebulous place. Many commentators would say that it's populated by firms with sales between $50 million and $500 million in sales, but other numbers are often used. Mitchell Berlin sive. Of course, 2M can always phone its ac count manager at Regional to request a tempo rary waiver or renegotiation of a covenant. For example, last year when new equipment pur chases threatened to reduce 2M's liquid assets on hand and push its working capital (cash plus accounts receivable) below the minimum level stipulated in the loan contract, 2M's owner called Regional. After a review of 2M's books and some further discussions to make sure that the fall in 2M's working capital was not due to other, more ominous causes, Regional offered a temporary waiver of the covenant.6 But renegotiations are not always easy. Some times, Regional has demanded an increase in the loan rate in exchange for a relaxation of the covenant. Sometimes, Regional has demanded an offsetting tightening of another covenant. For example, during the last negotiations, al though Regional allowed 2M's working capi tal to fall below the level usually considered prudent in the industry, it also demanded a reduction in the firm's debt-to-equity ratio. In fact, Regional and 2M have not always seen eye to eye about the risks of 2M's operating decisions, and the bank has not always agreed to contractual changes on any terms. Had Re gional viewed 2M's recent decline in working capital as too risky, negotiations could easily have turned out unsuccessfully for 2M.7 In this case, the firm might have been forced to post pone the equipment purchase or to search for another banker willing to provide funds (after duplicating Regional's investigation of the 6My account of renegotiations relies heavily on the ar ticles in "A Forum on the Effects of Violating Debt Cov enants," in the Accounting Review. 7Even when both the bank and the firm agree about the underlying riskiness of an operating decision, they may disagree about the desirability of the decision. As a credi tor with a fixed claim, the bank has a tendency to be espe cially wary of risky decisions, because it does not share in the high returns when the decision turns out especially well. 7 BUSINESS REVIEW firm's finances). Either outcome would have been costly for the firm. To Reduce Lender Power, Larger Firms Of ten Seek to D iversify T heir Sources of Funds...By diversifying the firm 's funding sources, 2M's owner feels that she would gain more discretion over production and invest ment decisions and also more bargaining power in negotiations with Regional. One possibility is that 2M could simply borrow from multiple banks, including Regional. Another possibil ity—which entails more fundamental changes in the ways that the firm does business—is that 2M could sell securities to the public in an ini tial public offering (IPO). 2M is now large enough to bear the costs of selling public secu rities, which include the substantial and ongo ing costs of providing information both to in vestors and to the SEC, as well as the fees paid to the und erw riting firm that brings the company's securities to market. Since 2M 's owner has been looking to diversify her per sonal portfolio by reducing her large stock hold ing in the firm, 2M elects an IPO. ...But Large Firms' Public Security Holders Continue to Value Bank Relationships. Al though the decision to sell public securities will ultimately weaken the intensity of 2M's rela tionship with Regional (indeed, this is one of the reasons for 2M's decision), the firm will con tinue to benefit from maintaining a lending re lationship. In fact, one of these benefits will be felt immediately. One of the enduring empirical puzzles in fi nancial economics is that stock sold in an IPO seems to be underpriced, in the sense that the initial buyers can turn around and resell the stock at a higher price. There is no consensus about why IPOs are underpriced, but most economists believe that it's related to investors' uncertainty about the quality of a firm new to public markets; thus, less uncertainty about the firm's prospects would reduce the amount of underpricing. This is just what 2M's relationship with Re 8 NOVEMBER/DECEMBER1996 gional appears to do. Otherwise suspicious in vestors act as if they view a prior borrowing relationship with a bank as good news about the firm, a type of Good Housekeeping Seal of Approval, which reduces their uncertainty about the firm's future prospects. 2M can rea sonably expect that its own stock will sell at a higher initial price than that of a similar firm that doesn't have an ongoing relationship with a bank. So, in 2M's case, the extent of underpricing is likely to be reduced, which is good for 2M, since the firm will get more funds from investors when it sells its securities.8 After 2M has gone public, its relationship with Regional will continue to affect the price of its public securities. This is true even though a firm with publicly traded securities must dis close a lot of information about its business affairs so that investors and analysts can form their own opinions and make their own fore casts about the firm's prospects. Investors will continue to react whenever 2M renews or rene gotiates its loan commitment with Regional. As long as the new contractual terms do not indi cate a worsening of 2M's financial situation— say, a higher rate than in the previous loan com mitment—2M's stock price will typically rise with the public announcement of the new loan contract. This positive stock-price reaction to an nouncements of bank loans and loan commit ments—an effect that has been found in study after study—stands in sharp contrast to the usually insignificant or negative effect of the announcement of a new public debt issue.9 8The evidence about bank relationships and IPOs can be found in two articles, one by Christopher James and Peggy Weir and another by Myron Slovin and John Young. 9The positive stock price effect when a loan agreement is announced is significant only when the number of banks lending to the firm is small (as discussed in the next sec tion). FEDERAL RESERVE BANK OF PHILADELPHIA For Better and For Worse: Three Lending Relationships Investors' willingness to pay more for the firm's stock suggests that they view the renewal of the loan relationship (on favorable terms) as good news about its future prospects, either because the bank's information about the firm's condi tion is superior to that of other investors—bank certification— or because the firm's stockholders believe that close supervision by the bank of the firm's affairs is likely to improve the firm's performance— bank monitoring. The positive effect of such an announcement has been found to be strongest when markets are uncertain about the firm's prospects—for example, when stock analysts have substantial disagreements about the firm's future earn ings— or if the firm's stock price has been low and investors have been pessimistic about the firm's earnings prospects. This effect has also been found to be strongest when the bank's cred ibility— as measured by its credit rating—is greatest. All of these findings support the idea that investors place a value on the lending re lationship.1 0 The story of 2M illustrates that the nature of the lending relationship changes over a firm's life-cycle. The tensions of exclusive lending relationships create powerful pressures for firms to diversify their funding sources when they become large enough. Further, the lend 10A thorough review of the literature on the stock-price effects of loan agreement announcements can be found in the article by Matthew Billet, Jon Garfinkel, and Mark Flannery. This article also performs an especially careful reexamination of prior findings. Notably, the authors call into question two earlier findings. Initial evidence indicated that only bank loans— and not other types of private debt— have positive announcement effects. Billet, Garfinkel, and Flannery summarize and add to the mounting evidence that all types of private debt have positive announcement effects. They also cast serious doubt on the prevailing be lief that announcement effects are significant only for re newals and renegotiations of loan agreements, but not for first-time agreements between firms and banks. Instead, they find that announcement effects are positive and sig nificant for both renewals and first-time agreements. Mitchell Berlin ing relationship itself eases the transition from exclusive borrowing from a single bank to di versified funding, especially borrowing on pub lic markets. But even when a firm secures funds from public securities markets, there is a pay off to the firm that maintains ties with its banker, because the banking relationship continues to convey information to investors. A FIRM'S ACCESS TO PUBLIC DEBT REDUCES THE BANK'S FLEXIBILITY While the empirical evidence says that banks play a continuing role in evaluating and moni toring firms with public securities—at least until the firm reaches a very large size—another aspect of the lending relationship seems to un dergo a fundamental change when the relation ship becomes less exclusive. The ease of rene gotiating bank loans, often seen as one of the hallmarks of lending relationships, appears to suffer. One piece of evidence that illustrates this loss of flexibility is that the positive stock-price ef fect of a loan announcement depends on there being a small number of lenders. Many firms borrow from a syndicate of banks: one bank negotiates the loan commitment agreement on behalf of a number of other banks, but all mem bers of the syndicate must ratify any adjust ments in the loan agreement. When the loan agreement involves a syndicate of more than three banks, the positive effect of the loan an nouncement on the firm's stock price disap pears.1 This finding makes sense because it is 1 more difficult to renegotiate loans with a syn dicate of banks than with just one or two banks; monitoring and controlling the firm through nThe empirical evidence on the stock-price effect of loan commitments by lending syndicates can be found in the article by Diana Preece and Donald Mullineaux. Their ar ticle considers— and rejects— a number of alternative ex planations for the insignificant stock-price effects of large syndicated loans. 9 BUSINESS REVIEW NOVEMBER/DECEMBER1996 covenants is much more valuable when contrac tual terms can be readily revised as new infor mation arrives and circumstances change. In effect, syndicated loan agreements are more like public debt—which is difficult to renegotiate— than a traditional bank loan. And when the firm actually has public debt—even just a little—bank debt is no longer so easy to negotiate when a firm is in finan cial distress. Consider Q C ontinuum C ast ings, which has sales of $250 million.1 Its 2 only bank is M ostly Derivatives Bancorp (M D B), w hich p ro vides Q with most of its short-term financ ing. In 1987 Q issued its first public debt, fol lowing many other middle market firms that had entered public debt markets for the first time. The debt was used to finance the purchase of a small HMO, a testament to Q's forwardlooking management, but a business outside Q's core market. This public debt represents only about 15 percent of Q's total debt financ ing, and MDB holds virtually all of the rest of Q's debt. For two years, sales of castings have been lagging while the HMO business has been booming. But since the HMO is only a small part of Q's businesses, it now appears that the firm will default on its loans to MDB unless it can somehow reduce its debt payments. Q has already entered negotiations with MDB, be cause its current ratio (working capital divided by total assets) has fallen below the minimum specified in its loan agreements, placing the company in technical default. As is common in loan contracts, a technical default must be rem edied within 60 days, or MDB has the right to demand immediate repayment of the loan. During negotiations, Q argues that the dol drums in the castings market are only tem porary and also notes that its HMO subsid iary is doing very well. The firm asks MDB to transform half of its short-term loans into long-term loans with lower face value and to exchange the re mainder for a substan tial share of Q's stock. These changes would have the effect of reduc ing Q's current interest payments and postpon ing payments to the bank to the future, which the firm is convinced will be brighter. MDB agrees that the long-term prospects in the castings market are reasonably favorable, but it makes a counterproposal. First, all of Q's public bondholders must exchange one-half of their bonds for stock. Second, the bank will al low Q to stretch out its short-term loan pay ments, but it will not reduce the face value of its debt. Third, MDB demands that Q sell off the HMO and use the proceeds from the sale to retire some of its bank debt. Finally, the bank demands a first lien on the machines used to produce castings, that is, Q's casting equipment will now serve as collateral for the bank's loans. The bank explains both its refusal to accept the company's offer and its own counteroffer as follows. The pain must be shared among all claimants, and it is not the bank's responsibil ity to bail out Q's bondholders. If the bank writes down the face value of the debt and re ceives stock in exchange, Q's bondholders will Even a small amount of public debt creates a conflict between the interests of the bank and those of the firm's bondholders. 12The story of Q is based on numerous articles, but it relies most heavily on a pair of significant papers by Chris topher James. These articles contain extensive bibliogra phies and good discussions of the previous empirical work on banks' role in debt renegotiations for financially dis tressed firms. 10 FEDERAL RESERVE BANK OF PHILADELPHIA For Better and For Worse: Three Lending Relationships be receiving substantial interest payments while the bank waits for Q's finances to improve enough to begin paying dividends. And in the worst possible case, if Q does not recover and enters bankruptcy, the bank's claim would be subordinate to those of bondholders. This means that the bondholders would be paid off, while the bank would be left with an equity share that may turn out to have little or no value. By forcing Q to sell its valuable HMO subsidiary to retire bank debt and by taking collateral in the castings business, MDB guar antees that it will recover at least some of its investment, even in this worst possible case. Although Q's management and its public bondholders feel that MDB is taking an unrea sonably harsh stance, they have little choice but to accept. As a result of these renegotiations with MDB, Q does avoid the bankruptcy courts, which usually eat up valuable resources like management time and attention, not to men tion expenses such as court and lawyers' fees. Yet, the firm has lost its prized jewel (the HMO), and Q's bondholders have been forced to shoul der a disproportionate share of the concessions. The first lesson of the story of Q is that even a small amount of public debt creates a conflict between the interests of the bank and those of the firm's bondholders. The source of this con flict is that the bondholders are the primary beneficiaries if the bank takes a conciliatory stance in debt renegotiations—for example, by taking equity in the firm or forgiving principal payments. The firm's public debt tends to harden its bank's bargaining position, as the bank makes sure that it does not bail out the firm's bondholders by making concessions. However, the second lesson is that Q's pri mary reliance on bank loans does ease negotia tions to avoid a costly bankruptcy. After all, it would have been much more difficult for Q to achieve an agreement with the bondholders alone. MDB is well informed about Q's finances because of its relationship with the firm, and one-on-one negotiations between two well-in Mitchell Berlin formed parties—Q's and MDB's managers— are likely to be better organized and less frac tious than negotiations with bondholders. Even though the bondholders realize that MDB's in terests and their own conflict, they also know that as Q's main creditor the bank stands to lose a lot if it permits the firm to continue opera tions and Q ultimately fails. MDB's willingness to renegotiate, rather than pull the plug and demand immediate repayment, signals to Q's bondholders the bank's informed belief that the firm is more valuable as an ongoing business. This makes them more likely to exchange their debt for stock.1 3 CONCLUSION The empirical literature of the last 10 years has uncovered some interesting lessons about the advantages and disadvantages of relation ship lending and about the ways that lending relationships change as competitive conditions facing a firm change. W here firms h ave limited financing choices—for example, small firms— relationship lending generates real benefits. Relationship lending is characterized both by close monitoring of the firm by the bank and by contractual flexibility. The possibility of long term lending relationships may make it easier for small, risky firms to borrow outside funds, but firms inevitably seek out more diversified funding sources when these become available. Indeed, a firm's prior relationship with a bank makes it easier for the firm to gain access to public securities markets, and even when the firm can issue public securities, bank relation ships continue to play a role. For all but the larg 13In his 1995 study, James also finds that bank equity participation in a debt restructuring is associated with su perior performance by the firm over the succeeding three years. This finding is tantalizing, but it is particularly diffi cult to disentangle the direction of causality. Did the bank take equity because of the firm's superior prospects, or did the firm prosper because the bank took equity? 11 BUSINESS REVIEW NOVEMBER/DECEMBER1996 est firms, banks continue to have an informa tional advantage that markets recognize. But diversification of funding sources severely lim its the bank's willingness to be flexible when firms enter financial distress, even when firms have only small amounts of public debt. None theless, a close relationship with a bank does increase the likelihood of successful renegotia tion when a firm enters financial distress. REFERENCES Berger, Allen N., and Gregory F. Udell, "Relationship Lending and Lines of Credit in Small Firm Finance," Journal of Business, 68,1995, pp. 351-81. Bhattacharya, Sudipto, and Anjan V. Thakor, "Contemporary Banking Theory," Journal of Financial Interme diation, 3 ,1993, pp. 2-50. Billet, Matthew T., Jon A. Garfinkel, and Mark J. Flannery, "The Effect of Lender Identity on a Borrowing Firm's Equity Return," Journal of Finance, 50,1995, pp. 699-718. "Forum on the Effects of Violating Debt Covenants," in The Accounting Review, 68, April 1993, pp. 219-303. James, Christopher M., "When Do Banks Take Equity? An Analysis of Bank Loan Restructurings and the Role of Public Debt," Review of Financial Studies, 8,1995, pp. 1209-34. James, Christopher M., "Bank Debt Restructurings and the Composition of Exchange Offers in Financial Distress," Journal of Finance, 51,1996, pp. 711-28. James, Christopher M., and Peggy Weir, "Borrowing Relationships, Intermediation, and the Cost of Issuing Public Securities," Journal of Financial Economics, 28,1990, pp. 149-71. Mester, Loretta J., "Banking and Commerce: A Dangerous Liaison?" Business Review, Federal Reserve Bank of Philadelphia, M ay/June 1992, pp. 17-29. Nakamura, Leonard I., "Recent Research in Commercial Banking: Information and Lending," Financial Markets, Institutions, and Instruments, 2,1993, pp. 73-88. Petersen, Mitchell, and Raghuram Rajan, "The Benefits of Lending Relationships: Evidence from Small Business Data," Journal of Finance, 49,1994, pp. 3-37. Petersen, Mitchell, and Raghuram Rajan, "The Effect of Credit Market Competition on Lending Relation ships," Quarterly Journal of Economics, 1995, pp. 407-43. Preece, Diana, and Donald J. Mullineaux, "Monitoring, Loan Renegotiability, and Firm Value: The Role of Lending Syndicates," Journal of Banking and Finance, 1996, pp. 577-94. Slovin, Myron, and John E. Young, "Bank Lending and Initial Public Offerings," Journal of Banking and Finance, 1990, pp. 729-40. 12 FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects of City and Suburban Job Growth u JL JL o w does the location of new jobs in a metropolitan area affect the suburban housing market? Economists expect job growth to in crease the demand for housing, and further more, they expect the increase in demand to be greater in communities near the new jobs than in more distant ones. Moreover, growth in jobs with higher wages should increase the demand for housing more than growth in jobs with lower wages. Increases in housing demand, in turn, put upward pressure on house prices and *Dick Voith is an economic advisor in the Research De partment of the Philadelphia Fed. Richard Voith* construction rates. The market response to an increase in housing demand, however, will de pend on how easily the supply can adjust to shifts in demand. Because of differences in proximity, wages, and housing supply, city employment growth may have dramatically different effects from suburban growth on house prices and construction rates across sub urban communities. These housing market ef fects provide a window through which we can evaluate the overall economic contributions of city and suburban job growth as well as insights into who benefits from city and suburban job growth. 13 BUSINESS REVIEW JOB GROWTH AT EMPLOYMENT CENTERS AND ON THE URBAN FRINGE Job growth in a metropolitan area may be widely dispersed geographically or it may oc cur in clusters. In its most extreme form, dis persed development is accommodated by wid ening the boundaries of the urbanized area rather than by increasing the density or num ber of jobs in areas already developed.1 Job growth in clusters, on the other hand, results in more intensive use of space in existing employ ment centers. Development at employment cen ters occurs both in the suburbs and in the city. We will refer to widely dispersed job growth that occurs primarily on the urban fringe as decentralized job growth and growth in clus ters as centralized growth. With decentralized development, agricul tural land is converted to commercial uses, and the demand for houses near the urban fringe increases. Because a great deal of open land is available, increases in housing demand are met by the construction of new houses. Price in creases for existing houses are limited by the cost of new construction. In economists' terms, housing supply is elastic on the urban fringe— shifts in demand cause small changes in price but large changes in the rate of construction. (See Prices, Construction Rates, and the Elasticity o f Supply.) While decentralized job growth tends to increase construction rates on the ur ban fringe, it may have little effect on either prices or construction in neighborhoods away from the fringe because the number of jobs ac cessible to these communities is essentially un changed. Finally, if decentralized growth occurs at the expense of jobs in existing employment M etropolitan areas generally consist of one or more counties. Within the boundaries of the metropolitan areas, there are usually one or more central cities, suburbs sur rounding the central cities, and land used for agriculture beyond the suburbs. Taken together, the central cities and the suburbs constitute the urbanized area, and the agricul tural land adjacent to the urbanized area is the urban fringe. Digitized 14 FRASER for NOVEMBER/DECEMBER1996 centers, it may reduce the demand for housing near these centers. This would tend to reduce housing prices and, to a lesser extent, the num ber of houses in these areas. Centralized job growth, on the other hand, increases the demand for housing in commu nities with easy access to the employment cen ter. Because these neighborhoods tend to be densely developed, it is difficult to construct new housing. Housing supply in these commu nities is inelastic—increases in demand tend to drive house prices up, with little or no impact on construction. Centralized em ploym ent growth, in theory, should have price impacts that decline with distance from the center. Dis tant communities may even experience a de cline in demand if centralized job growth is a result of shifts in employment from the urban fringe to employment centers. Basic models of urban economies highlight the role of centralized production and the im portance of proximity to these employment cen ters for property values.2 The earliest models assumed that all production occurred in the center of the area and everyone commuted to the central area to work. Workers seeking to avoid high commuting costs bid up prices near the employment center so that the value of land fell as distance from the center increased. Rather than maintaining the unrealistic as sumption that all production occurs in the cen ter, more recent models consider both central ized and decentralized employment.3 A basic assumption of many of these models is that firms locate near one another because doing so has economic advantages: workers are more 2See the articles by Edwin Mills and Richard Muth for a discussion of equilibrium models of monocentric urban economies, that is, economies in which there is a single fo cal point where production occurs. 3The papers by Arthur Sullivan, Jan Brueckner, and Michelle White are good examples of monocentric urban models with decentralized employment. FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects of City and Suburban Job Growth Richard Voith Prices, Construction Rates, and the Elasticity of Supply Markets adjust to achieve equilibrium in two ways: by changing the price of goods traded and by chang ing the quantity of goods traded. Given a shift in housing demand, the extent to which the adjustment will be accomplished by price changes or by new house construction will depend on the elasticity of supply. Economists define the elasticity of supply as the percentage change in quantity divided by the percentage change in price. If the change in price is greater than the change in quantity, supply is said to be inelastic (Figure A); if the change in quantity is greater than the change in price, supply is said to be elastic (Figure B). FIGURE A FIGURE B Figure A shows the equilibrium price, p*, and quantity, q*, with demand curve D and supply curve S. A shift in demand from D to D' results in a shift in quantities to q'. The supply curve is almost vertical, indicating that an increase in price changes quantity supplied very little; hence, supply is inelastic. Thus the new equilibrium is achieved with a greater shift in price than in quantity. If supply were perfectly inelastic, the quantity supplied would not change at all with a change in demand, and all adjustments would be achieved by changing prices. Inelastic supply corresponds to the housing market conditions in older, densely developed neighborhoods. Figure B shows a similar diagram, but this time with a flatter supply curve. In this case, the shift in demand results in a much greater shift in quantity than in price. The flatter supply curve indicates that small changes in price induce large changes in quantity; hence, supply is elastic. If the supply curve were perfectly flat, shifts in demand would not affect price; only quantity would change. Elastic supply corre sponds to housing market conditions in communities on the urban fringe. productive in areas where a lot of economic because their commuting costs are lower. These activity occurs. Economists call these advan expanded models, which provide a rationale for tages "agglomeration economies." While ag both centralized and decentralized employglomeration economies induce many firms to concentrate in commercial centers, other firms 4It is likely that firms choosing central locations are ones still choose d ecen tralized lo catio n s.4* that can benefit the most from agglomeration economies, Noncentrally located firms offer lower wages, while those choosing decentralized locations are in busi but some workers choose to work at these firms nesses that do not benefit from agglomeration. 15 BUSINESS REVIEW merit, still predict that land prices decline with distance from employment centers. There is a long tradition of estimating the relationship between housing prices and dis tance to high-productivity employment cen ters.5 Changes in this relationship are of inter est because they provide insight into the change in the benefits of agglomeration over time.6 There is little analysis, however, of the conse quences of shifts in production between cen tralized and decentralized locations. Such analysis could provide insight into who ben efits and by how much from the change in land values associated with centralized or decentral ized job grov/th. Homeowners and developers may have di vergent interests in the pattern of job growth. Employment growth in existing centers, such as the central business district (CBD), is likely to enhance the value of existing houses near job centers. Decentralized growth at the urban fringe is likely to result in shifts of agricultural land to residential use, but it will have little or no impact on the price of existing houses. De- 5See the articles by Paul Waddell, Brian Berry, and Irv ing Hoch; E. Heikkila and colleagues; and John McDonald and Daniel McMillen for analyses of land value gradients in a polycentric context. In a monocentric setting, there is mixed evidence on the relationship between land values and distance from the central business district. Analyses by M. Cropper and Peter Gordon; the article by E. Heikkila and colleagues; and my 1991 article find either a positive or insignificant relationship between land prices and dis tance. Other studies, such as the one by Rena Sivitanidou in 1996 and my 1993 article, find the expected negative re lationship. See the article by J. Jackson for a review of ear lier studies. 6While the relationship between land prices and dis tance should reflect the value of agglomeration, research ers have generally focused on the relationship between population density and distance to evaluate the relative importance of basic economic factors such as technology and income compared with urban problems such as crime to the process of decentralization. See the article by Edwin Mills and Peter Mieszkowski for a review of this literature. 16 NOVEMBER/DECEMBER1996 velopers and owners of agricultural land are thus the primary beneficiaries of the demand shift associated with decentralized growth. Centralized and decentralized employment growth also have a potentially different impact on the total value of land. To the extent that cen tralized employment is more productive than decentralized employment, it will have a larger impact on total land value. Of course, the dis tribution of jobs is a result of choices by indi vidual firms, which can best decide where their workers are most productive. In a perfectly competitive market with no spillovers, the lo cation decisions of individual firms should re sult in the most efficient total production and ultimately the highest total value for residen tial land. But the basic notion of agglomera tion economies is that an individual firm's choice of location affects other firms.7 One firm's decision to move out of the CBD has negative consequences for the remaining firms, which lose some of the benefits of concentra tion. The firm choosing to leave does not have to pay the costs imposed on other firms. Thus, private incentives may result in an inefficiently rapid pace of decentralization and lead to less efficient production and lower total land value. EMPLOYMENT GROWTH AND HOUSING IN GREATER PHILADELPHIA In a recent paper I estimated the effects of city and suburban job growth on the housing market in Montgomery County, Pennsylvania, in the Philadelphia suburbs.8 In my study, sub urban employment growth included job growth 7A gglom eration econom ies are like a local public good—everyone choosing to locate in the community ben efits from it. Standard microeconomic theory suggests that the competitive market will result in an inefficiently low level of local public goods. 8My 1996 paper provides a complete description of the analysis. FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects of City and Suburban Job Growth Richard Voith in Montgomery County and three neighboring mine the rate of new construction annually for suburban counties, and city em ploym ent each census tract and the average real appre growth included only that in the city of Phila ciation and construction rates over the sample 0 delphia (Figure 1). While city and suburban job period (Figures 2a and 2b).1 The figures show growth do not exactly correspond to central that prices and construction rates vary a great ized and decentralized job growth, we expect deal over time. This variation in appreciation that the impact of city growth should be more like that of centralized growth.9 To the extent that suburban job growth is widely dispersed 10The data come from the 1988,1994, and 1995 tax-as and occurs substantially on the urban fringe, sessment files of Montgomery County. The appreciation we expect the effects of suburban growth to be rates are based on 1990 constant dollars. more like those of de centralized growth. FIGURE 1 The data for the Philadelphia and the Pennsylvania Suburbs study in clu d e m ore than 88,000 sales of single-family detached houses from 1972-95. Detailed information, such as sale price, year of sale, characteristics, size, and location, is av ailab le for each Bucks house. Since the data Salford< set also includes infor mation on virtually all houses in Montgomery County, their construc Montgomery tion date, and their cen sus tract, we can deter 9 Because of the high den sity of both employment and population throughout the city of Philadelphia, all city job growth is thought of as centralized growth, which in creases the intensity of land use in the existing communi ties. This is not necessarily true for all central cities in the U nited States, since som e central cities have significant amounts of agricultural land within their boundaries. Chester Delaware 17 NOVEMBER/DECEMBER1996 BUSINESS REVIEW FIGURE 2 Montgomery County Single-Family Detached Houses A. Price Appreciation Percent B. Growth in Housing Stock Percent 18 and construction rates is, in part, related to employ ment growth. Moreover, ap p reciatio n and co n stru ction rates vary greatly across communi ties, and these differences are linked to the patterns of em ployment growth. There are substantial dif ferences in the movements of city and suburban job growth (Figure 3). While suburban growth rates are almost always larger than city grow th rates (city growth rates are predomi nantly negative), the dif ference in city and subur ban growth rates varies su b stan tially over ex tended periods. To evaluate the effect of city and suburban em ployment growth on sub urban house prices and construction rates, we con structed a statistical model to take into account as m any factors affectin g house prices as possible, including the characteris tics of the house such as its age, number of rooms, and number of bathrooms, as well as the size of its lot, its location, and the neigh borhood.1 After control 1 ling for these influences on price, we evaluated the ef fects of city and suburban employment on price and whether the effects differ systematically with acces sibility to the city. Similar FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects of City and Suburban Job Growth Richard Voith FIGURE 3 analyses were undertaken Employment Growth: with regard to factors that affect construction rates. Suburban Rate Minus City Rate Employment Growth (3-Month Moving Average) and House Prices. Our statistical models of house prices suggest that city and subu rban grow th have very different effects on the suburban housing m arket, and the effects vary dramatically across suburban com m unities. Our sim p lest m odel, which is designed to cap ture the overall effect of city and suburban em p loym ent grow th on house prices, reveals that 72 74 76 78 80 82 84 86 88 90 92 94 city employment growth Year has a positive effect on suburban house prices. Our estimates indicate accessibility to the city, either by commuter train that an increase in city job growth of 1 percent or by highway. For every 1 percentage point of age point raises average suburban house val growth in city employment, communities with ues slightly more than $1000.1 Suburban em commuter rail service enjoy an increase in house 2 ployment growth, on the other hand, has vir values that is more than $1500 greater than com tually no effect on average house prices. The munities without train service.1 (The reverse 3 estimated average effects, however, mask large is true as well: when city employment shrinks differences in effects across communities. 1 percentage point, the value of these houses More complex statistical models allow the falls.) effects of city and suburban em ploym ent The differential effects of accessibility and growth to vary depending on the community's employment are even more pronounced when we look at highway commuting times.1 The 4 effects of an increase of 1 percentage point in 11A common technique used in housing economics is hedonic regression. In this type of analysis, housing prices are regressed on housing and neighborhood traits. The estimated coefficients represent the "price" of individual housing traits such as number of rooms, number of bath rooms, and so forth. Housing traits and their estimated prices explain much of the variation in house prices. 12This estimate is based on Table 3, column 3, of my 1996 article. Based on transactions in the period from 199095, the average house price was $166,900, which implies that a 1 percentage point increase in city growth increases average suburban house values about 0.6 percent. 13These estimates are based on the model shown in Table 4, column 1, of my 1996 article. 14Again, these estimates are based on the model shown in Table 4, column 1, of my 1996 article. The effects are for a typical house in an average neighborhood. Because 42 percent of houses are in communities with train service, these effects include 42 percent of the estimated effect of employment growth on house values in communities with train service. 19 BUSINESS REVIEW NOVEMBER/DECEMBER1996 FIGURE 4 growth to have a negative effect on distant suburbs, people must be choosing to live closer to the CBD The effects vary by distance from the CBD. for reasons other than commuting. Perhaps city Change in House Price ($) em ploym ent growth is correlated w ith im proved, regionally valued amenities located in the city, which induce more people to choose locations that are closer in.1 5 The effects of subur ban job growth on house prices are relatively small. For a rise of 1 percentage point in suburban grow th, com m unities near the city suffer price declines in the range of Commute Time from House to CBD (minutes) $1300, and prices in dis tant com m unities in *The city effects are for a 1-percentage-point increase in city job growth, holding crease a scant $300.The suburban employment growth constant. Similarly, the suburban effects are for a 1difference across commu percentage-point increase in suburban employment, holding city job growth constant. nities is a relatively small $1600. The price decline either city or suburban employment growth in communities near the city may reflect other vary with commuting time to the CBD ( Figure factors that both lower the attractiveness of the 4). The vertical axis of Figure 4 shows the effect city and increase job growth in the suburbs. This on prices, and the horizontal axis gives the dis explanation is supported by the finding that tance from the CBD. For the communities clos suburban job growth has no differential impact est to the CBD— about a 20-minute commute— on communities with or without train service 6 a rise in city employment growth of 1 percent to the CBD.1 * In summary, our estimates imply that—at age point increases house values more than $5900. The same amount of growth reduces house values in the most distant communities about $1300. Thus, the difference in the impact 15The estimates give the effects of city growth, holding of city job growth on house prices across com the rate of suburban growth constant. Thus city employ munities is about $7200. ment reduces house values in distant suburbs even though The negative effect of city job growth on more city growth does not change the rate of suburban growth. distant communities implies that employment 16It is not surprising that suburban growth has no dif growth must be correlated with other factors ferential effect on communities with or without train ser that affect the relative attractiveness of the city vice, since train service adds little to accessibility to subur and suburbs. Note that for city employment ban jobs. City and Suburban Job Growth and House Prices* 20 FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects o f City and Suburban Job Growth least in the Philadelphia area—city employment growth has a significant, positive impact on suburban house prices while suburban employ ment growth, on average, does not. City em ployment growth has a strong differential ef fect on suburban house prices in communities with train service while suburban employment growth does not. While both city and subur ban employment growth affect communities differently according to their distance from the CBD, the effects are much larger for city than for suburban employment growth. In general, these findings are consistent with the idea that the supply of housing is inelastic in older com munities near the city and elastic near the ur ban fringe. Employment Growth and Housing Con struction. Markets adjust to changes in demand by shifting not only prices but also quantity. Quantity adjustments in response to new job creation are affected by other factors such as the availability of land (density) and mortgage interest rates. Therefore, we take into consider ation the effects of density and mortgage inter est rates in estimating the effects of job growth on housing construction.1 Density is a good 7 measure of the potential that a community has for new residential development. Communities with high density should have low construc tion rates because little open space is available for new construction.1 Mortgage interest rates 8 affect both the cost of financing construction for developers and the cost of financing purchases by home buyers. On average, the annual rate of construction 17In our statistical model, density has a very large nega tive impact on construction rates. Mortgage rates also have a negative, statistically significant effect. 18In a given year, an average of 42 percent of all census tracts have no new construction. The fact that observed construction rates are frequently zero necessitates the use of the Tobit procedure to estimate the construction mod els. Richard Voith for new, single-family detached housing in Montgomery County is about 1 percent of the total housing stock. City and suburban employ m ent grow th have op posite effects on countywide construction rates. We estimate that an increase of 1 percentage point in city job growth reduces the construction rate a little more than 0.1 percent.1 That is, the average 9 annual construction rate falls from 1 percent to about 0.9 percent. Suburban growth increases the average construction rate by a similar mag nitude: an increase of 1 percentage point in sub urban job growth raised the construction rate to almost 1.09 percent. As with price effects, there are large differ ences in the effects of job growth on construc tion rates among suburban locations.20* The impact of city employment growth on subur ban construction rates is an increase of 0.03 per cent in communities within a 20-minute com mute, but in the most distant communities, city employment growth reduces construction 0.27 percent (Figure 5). Thus, a healthy city economy reduces the relative attractiveness of the most distant communities. The effects of suburban job growth also differ by location. An increase of 1 percentage point in suburban employment growth raises the construction rate in distant communities 0.21 percent but has little effect on close-in communities. This suggests that new suburban jobs are truly decentralized; that is, they occur primarily at the fringe of the metro politan area. SHOULD WE CARE ABOUT THE GEOGRAPHY OF JOB GROWTH? There are two broad reasons why we should care where growth occurs: economic efficiency 19A11 estimates in this section are based on Table 6, col umn 1, in my 1996 article. 20Unlike the case with price effects, there were no dif ferential effects across communities with and without train service for either city or suburban employment growth. 21 NOVEMBER/DECEMBER1996 BUSINESS REVIEW FIGURE 5 w ill be w illing to pay more to locate there.2 1 City and Suburban Job Growth and Therefore, one way to the Rate of House Construction* evaluate the overall effi The effects vary by distance from the CBD. ciency of city and subur ban growth is to compute Growth of Housing Stock (%) which form of growth contributes more to the total value of land. The data in our study were limited to one suburban county, so we cannot evaluate the effect of job grow th on the overall Philadelphia metropoli tan area. But we can evaluate the effects of city and suburban job growth on the total value of resi dential land in M ont gomery County.2 2 There are two e le ments in the calculation of the impact on residen *The city effects are for a 1-percentage-point increase in city job growth, holding suburban employment growth constant. Similarly, the suburban effects are for a 1tial land value from any percentage-point increase in suburban employment, holding city job growth constant. job growth: the change in values for existin g and distribution of economic impacts. Ques houses and the change in property value asso tions about efficiency focus on which type of ciated with the construction of new houses.2 * 3 employment growth generates the greatest out First, consider the effects of city employment put for the region. Questions about distribution, growth. The house-price models imply that an on the other hand, focus on who benefits from increase of 1 percentage point in city employ city and suburban employment growth. When evaluating the efficiency and distributional is sues associated with city and suburban job 22This analysis is further limited to the effects on the growth, we need to combine the findings from value of single-family detached dwellings. the price and quantity sides of the housing 23Calculating the change in residential land values re market to determine the total impact. sulting from new construction is complicated by the fact Efficiency. When the average productivity that, ideally, we would like to know the difference between in a metropolitan area is high, people and firms the value of land before and after new construction. We do 21Of course, there are many other factors that affect people's choice of where to live, but for similarly attractive metro areas, higher productivity should increase land value. 22 not have a good measure of the value of vacant land or of the value of land after construction. We assume that the value of vacant land is zero and the value of newly devel oped land is equal to the average tract price, including housing. Thus, our estimate of the impact on land value represents an upper bound. FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects of City and Suburban Job Growth ment growth raises the aggregate value of resi dential property 0.58 percent.24 Recall that an increase of 1 percentage point in city job growth reduces the rate of construction 0.1 percent, that is, the rate of construction falls from 1 percent of the total housing stock per year to 0.9 per cent per year. This translates into a decrease in value of 0.1 percent from what the county would have experienced otherwise. The net impact of city job growth, therefore, is an in crease in residential value of 0.48 percent. The positive effect on price far outweighs the nega tive impact of city job growth on suburban con struction. Next, consider the effects of subur ban job growth on residential values. Suburban growth has virtually no impact on prices, but it does have a positive impact of 0.09 percent on construction. Thus, the effect of employment grow th on constru ction causes suburban growth to have an overall positive impact on residential value, but the magnitude of the over all impact is about one-fifth the impact of city job growth. If there are no differences in productivity between centralized and decentralized growth, the impacts of city and suburban growth should be similar in magnitude, so the above differ ence in the total impact implies that centralized employment tends to be more productive than decentralized employment. This immediately raises the question of why suburban employ ment is growing while city employment is de clining, and this question has essentially two possible answers. First, there may be negative spillovers from the decentralization of jobs. The choice by individual firms to leave the city may have negative effects on the remaining firms, which in turn may induce additional firms to leave. As we discussed earlier, this process may lead to an inefficiently rapid pace of decentrali- Richard Voith zation. Second, the higher compensation for jobs in Philadelphia may no longer reflect the true productivity differential between city and suburban employment, and thus we are sim ply in a transitional state in which relative wages between the city and suburbs are still adjusting. Because wages have not completely adjusted, jobs are moving out of the city. Of course, this second explanation can be a result of either technological forces favoring decen tralized production or an inefficient loss of ag glomeration economies. Distribution. While the differences in the impact of city and suburban job growth on the total value of residential real estate are signifi cant, the differences across communities are even more dramatic. These differences are il lustrated by the estimated effects of city and suburban job growth for two Montgomery County communities: Narberth and Salford. Narberth is a small, old, relatively dense com munity that lies 24 minutes by highway from the Philadelphia CBD. It also has train service. Salford is a community nearer the urban fringe, with low population density and no train ser vice, and getting to the Philadelphia CBD en tails a 77-minute commute. We estimate that an increase of 1 percentage point in city job growth will increase the value of residences in Narberth 3.15 percent, but house values in Narberth will suffer a decrease of 0.47 percent when suburban job growth in creases 1 percentage point.25 If we use an aver age house price of $196,300 in Narberth, job growth of 1 percentage point in the city would increase the value of a house $6180; similar job growth in the suburbs would reduce its value $920.2 On the other hand, house values in 6 25These estimates include the value of both price and construction impacts. 24See my 1996 article for details of this calculation. These calculations assume that the new construction is valued at the mean value of houses in the county. 26The average sales price is based on transactions oc curring in the period 1990-95. 23 NOVEMBER/DECEMBER1996 BUSINESS REVIEW Salford would suffer a decline of 1.38 percent if city jobs grow 1 percentage point but would gain 0.39 percent from sim ilar suburban growth. Since the average house value in Salford was $133,300, these estimates translate into a loss of $1840 from city growth and a gain of $520 from suburban growth. The finding that job growth can negatively affect total value in distant communities—city growth has a nega tive effect on Salford, and suburban growth has a negative effect on Narberth—indicates that employment growth in a community tends to make communities more attractive for reasons other than simply the availability of jobs. For example, communities experiencing job growth may be in a better position to provide quality education, security, and other amenities to their residents. SUMMARY While our analysis of the relationship be tween the housing market and employment growth indicates that employment growth in creases the value of real estate assets, all subur ban residents do not share equally in the in crease. Older, developed suburbs and suburban fringe communities do not have common in terests, at least in terms of the patterns of eco nomic growth and their effect on the housing market. Decentralized job growth increases the value of land on the urban fringe, and owners of agricultural land and developers are the prime beneficiaries. Centralized job growth, on the other hand, enhances property values in existing communities, and unfortunately, a de cline in centralized jobs reduces the value of these properties. REFERENCES Brueckner, Jan K. "A Model of Non-Central Production in a Monocentric City," Journal of Urban Economics, 6, 1979, pp. 444-630. Cropper, M.L. and Peter L. Gordon. "Wasteful Commuting: A Reexamination," Journal of Urban Economics, 29, 1991, pp. 2-13. Heikkila, E.J., Peter L. Gordon, J. I. Kim, et al. "What Happened to the CBD-Dislance Gradient? Values in a Polycentric City," Environment and Planning A, 21,1989, pp. 221-32. Jackson, Jerry R. "Intraurban Variation in the Price of Housing," Journal of Urban Economics, 6, 1979 pp. 46479. McDonald, John F., and Daniel P. McMillen. "Employment Subcenters and Land Values in a Polycentric Urban Area—The Case of Chicago," Environment and Planning A, 22,1990, pp. 1561-74. Mills, Edwin S. "An Aggregative Model of Resource Allocation in a Metropolitan Area," American Economic Review 47,1967 pp. 197-210. Mills, Edwin S., and Peter Mieszkowski. "The Causes of Metropolitan Suburbanization," Journal of Eco nomic Perspectives, 7 , 1993, pp. 135-47. Muth, Richard F. Cities and Housing: The Spatial Pattern of Urban Residential Land Use. Chicago: University of Chicago Press, 1969. 24 FEDERAL RESERVE BANK OF PHILADELPHIA The Suburban Housing Market: The Effects of City and Suburban job Growth Richard Voith Sivitanidou, Rena. "Are Service Agglomerations Weakening? Evidence from Metropolitan Los Angeles," mimeo, School of Urban and Regional Planning, University of Southern California, 1996. Sullivan, Arthur M. "A General Equilibrium Model with Agglomerative Economies and Decentralized Em ployment," Journal of Urban Economics, 20,1986, pp. 55-74. Voith, Richard P. "Transportation, Sorting, and House Values," Journal of the American Real Estate and Urban Economics Association, 19,1991, pp. 117-37. Voith, Richard P. "Changing Capitalization of CBD-Oriented Transportation Systems: Evidence from Phila delphia, 1970-1988," Journal of Urban Economics, 33,1993, pp. 361-76. Voith, Richard P. "The Suburban Housing Market: Effects of City and Suburban Employment Growth," Federal Reserve Bank of Philadelphia Working Paper 96-15. Waddell, Paul, Brian Berry, and Irving Hoch. "Residential Property Values in a Multinodal Urban Area: New Evidence on the Implicit Price of Location," Journal of Real Estate Finance and Economics, 7,1993, pp. 117-41. White, Michelle J. "Location Choice and Commuting Behavior in Cities with Decentralized Employment," Journal of Urban Economics, 24,1988, pp. 129-52. Yinger, John. "City and Suburb: Urban Models with More than One Employment Center," Journal of Urban Economics, 31,1993, pp. 181-205. 25 We have a new We’ve also changed our look and made it easier for you to get around our web site. Now you can more easily find: • • • • • • • • • Information about the Philadelphia Fed Economic data and surveys Bank publications and speeches Consumer information Material on community affairs and CRA Information for investors in government securities Media releases Answers to your questions about the Fed Links to other Federal Reserve sites and other sources of economic/financial information We even have a page just for our depository institutions. There, DFIs can locate information on regulatory issues and the many services we provide to banks, thrifts and credit unions. And, as always, our web site offers researchers access to historical economic data. Researchers * Analysts * Depository institutions * Local http://fraser.stlouisfed.org/ * Federal Reserve Bank of St. Louis Consumers * Investors * Educators * Community groups * Medi 1996 INDEX J anuary /February Lee E. Ohanian, "When the Bubble Bursts: Psychology or Fundamentals?" Keith Sill, "The Cyclical Volatility of Interest Rates" March/April Richard Voith, "Central City Decline: Regional or Neighborhood Solutions?" Gerald A. Carlino and Robert H. DeFina, "Does Monetary Policy Have Differential Regional Effects?" May/June Theodore M. Crone and Kevin J. Babyak, "Looking Ahead: Leading Indexes for Pennsylvania and New Jersey" Dean Croushore, "Inflation Forecasts: How Good Are They?" July/August Loretta J. Mester, "Repealing Glass-Steagall: The Past Points the Way to the Future" Gregory P. Hopper, "Value at Risk: A New Methodology for Measuring Portfolio Risk" September/October Satyajit Chatterjee, "Taxes, Homeownership, and the Allocation of Residential Real Estate Risks" Timothy Schiller, "The Travel Market in the United States and the Third District" November/December Mitchell Berlin, "For Better and For Worse: Three Lending Relationships" Richard Voith, "How Does City and Suburban Employment Growth Affect the Suburban Housing Market?" FEDERAL RESERVE BANK OF PHILADELPHIA Business Review Ten Independence Mall, Philadelphia, PA 19106-1574