Corporate Real Estate Sales and Agency Costs of Managerial Discretion



Similar documents
DO LOSS FIRMS MANAGE EARNINGS AROUND SEASONED EQUITY OFFERINGS?

Can Auto Liability Insurance Purchases Signal Risk Attitude?

Sulaiman Mouselli Damascus University, Damascus, Syria. and. Khaled Hussainey* Stirling University, Stirling, UK

Two Faces of Intra-Industry Information Transfers: Evidence from Management Earnings and Revenue Forecasts

STAMP DUTY ON SHARES AND ITS EFFECT ON SHARE PRICES

Day-of-the-Week Trading Patterns of Individual and Institutional Investors

The impact of hard discount control mechanism on the discount volatility of UK closed-end funds

The Investor Recognition Hypothesis:

Management Quality, Financial and Investment Policies, and. Asymmetric Information

New evidence of the impact of dividend taxation and on the identity of the marginal investor

Management Quality and Equity Issue Characteristics: A Comparison of SEOs and IPOs

The Probability of Informed Trading and the Performance of Stock in an Order-Driven Market

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*

Journal of Empirical Finance

An Alternative Way to Measure Private Equity Performance

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts

Gender differences in revealed risk taking: evidence from mutual fund investors

This study examines whether the framing mode (narrow versus broad) influences the stock investment decisions

The Impact of Stock Index Futures Trading on Daily Returns Seasonality: A Multicountry Study

Macro Factors and Volatility of Treasury Bond Returns

How To Calculate The Accountng Perod Of Nequalty

Do stock prices underreact to SEO announcements? Evidence from SEO Valuation

Bank Credit Conditions and their Influence on Productivity Growth: Company-level Evidence

Traditional versus Online Courses, Efforts, and Learning Performance

The Journal of Applied Business Research January/February 2010 Volume 26, Number 1

Online Appendix Supplemental Material for Market Microstructure Invariance: Empirical Hypotheses

WORKING PAPER SERIES TAKING STOCK: MONETARY POLICY TRANSMISSION TO EQUITY MARKETS NO. 354 / MAY by Michael Ehrmann and Marcel Fratzscher

Solution: Let i = 10% and d = 5%. By definition, the respective forces of interest on funds A and B are. i 1 + it. S A (t) = d (1 dt) 2 1. = d 1 dt.

THE DETERMINANTS OF THE TUNISIAN BANKING INDUSTRY PROFITABILITY: PANEL EVIDENCE

SPECIALIZED DAY TRADING - A NEW VIEW ON AN OLD GAME

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

The DAX and the Dollar: The Economic Exchange Rate Exposure of German Corporations

Informational Content of Option Trading on Acquirer Announcement Return * National Chengchi University. The University of Hong Kong.

Whose Private Benefits of Control. Owners or Managers?

Construction Rules for Morningstar Canada Target Dividend Index SM

Criminal Justice System on Crime *

Accounting Discretion of Banks During a Financial Crisis

The Short-term and Long-term Market

Chapter 15: Debt and Taxes

Student Performance in Online Quizzes as a Function of Time in Undergraduate Financial Management Courses

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

World currency options market efficiency

The timing ability of hybrid funds of funds

Clay House Case Study and Comparison of Two Behemoths ofEC term

Small and medium-sized enterprises, bank relationship strength, and the use of venture capital

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall

Empirical Evidence of Trade Credit Uses of Brazilian Publicly Listed Companies

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Financial Instability and Life Insurance Demand + Mahito Okura *

DEFINING %COMPLETE IN MICROSOFT PROJECT

THE RELATIONSHIP BETWEEN FINANCING POLICY AND FINANCIAL PERFORMANCE IN THE BRAZILIAN TEXTILE INDUSTRY

Transaction Costs and Strategic Trading of German Investment Management Firms: Empirical Evidence from European Stock Markets

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy

THE IMPLIED VOLATILITY OF ETF AND INDEX OPTIONS

A Model of Private Equity Fund Compensation

Hot and easy in Florida: The case of economics professors

International Commodity Prices and the Australian Stock Market

Understanding the Impact of Marketing Actions in Traditional Channels on the Internet: Evidence from a Large Scale Field Experiment

Tax Avoidance Strategies in (Probably) Loss-making

To manage leave, meeting institutional requirements and treating individual staff members fairly and consistently.

Corporate Governance Compliance and the Effects to Capital Structure in Malaysia

Is There A Tradeoff between Employer-Provided Health Insurance and Wages?

The Choice of Direct Dealing or Electronic Brokerage in Foreign Exchange Trading

Foreign Direct Investment in a World of Multiple Taxes

Risk-Adjusted Performance: A two-model Approach Application in Amman Stock Exchange

Small Business Loan Turndowns, Personal Wealth and Discrimination

Research Division Federal Reserve Bank of St. Louis Working Paper Series

Working Paper The determinants of the flow of funds of managed portfolios: mutual funds versus pension funds

Beating the Odds: Arbitrage and Wining Strategies in the Football Betting Market

HARVARD John M. Olin Center for Law, Economics, and Business

Kiel Institute for World Economics Duesternbrooker Weg Kiel (Germany) Kiel Working Paper No. 1120

THE EFFECT OF PREPAYMENT PENALTIES ON THE PRICING OF SUBPRIME MORTGAGES

Analysis of Premium Liabilities for Australian Lines of Business

Efficient Project Portfolio as a tool for Enterprise Risk Management

Chapter 15 Debt and Taxes

Momentum Trading, Mean Reversal and Overreaction in Chinese Stock Market *

1.1 The University may award Higher Doctorate degrees as specified from time-to-time in UPR AS11 1.

ESTIMATING THE MARKET VALUE OF FRANKING CREDITS: EMPIRICAL EVIDENCE FROM AUSTRALIA

A Multistage Model of Loans and the Role of Relationships

Analyzing Search Engine Advertising: Firm Behavior and Cross-Selling in Electronic Markets

Are Women More Likely to Seek Advice than Men? Evidence from the Boardroom

Are Women Better Loan Officers?

How To Get A Tax Refund On A Retirement Account

Trade Adjustment and Productivity in Large Crises. Online Appendix May Appendix A: Derivation of Equations for Productivity

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

Heterogeneous Paths Through College: Detailed Patterns and Relationships with Graduation and Earnings

! # %& ( ) +,../ # 5##&.6 7% 8 # #...

An Interest-Oriented Network Evolution Mechanism for Online Communities

High Correlation between Net Promoter Score and the Development of Consumers' Willingness to Pay (Empirical Evidence from European Mobile Markets)

Overview of monitoring and evaluation

AN EMPIRICAL INVESTIGATION OF IPO UNDERPRICING IN CHINA

Start me up: The Effectiveness of a Self-Employment Programme for Needy Unemployed People in Germany*

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

Buy-side Analysts, Sell-side Analysts and Private Information Production Activities

Using Series to Analyze Financial Situations: Present Value

Discount Rate for Workout Recoveries: An Empirical Study*

Vasicek s Model of Distribution of Losses in a Large, Homogeneous Portfolio

Creditor Rights Protection, Ultimate Ownership and the Debt Financing Costs and Ratings: International Evidence

Number of Levels Cumulative Annual operating Income per year construction costs costs ($) ($) ($) 1 600,000 35, , ,200,000 60, ,000

Transcription:

Corporate Real Estate Sales and Agency Costs of Manageral Dscreton Mng-Long Lee * Department of Fnance Natonal Yunln Unversty of Scence & Technology Yunln, Tawan Mng-Te Lee Department of Accountng Tamkang Unversty Tape, Tawan * Correspondence: Mng-Long Lee, Department of Fnance, College of Management, Natonal Yunln Unversty of Scence and Technology, 123 Secton 3, Unversty Road, Doulu, Yunln, Tawan 640. Phone: (05) 534-2601 ext. 5338, Fax: (05) 531-2079, E-mal: leemng@yuntech.edu.tw. Mng-Long Lee would lke to acknowledge support from Tawan Natonal Scence Councl grant NSC 92-2416-H-224-010-.

Corporate Real Estate Sales and Agency Costs of Manageral Dscreton Abstract Ths study contrbutes to the lterature on corporate real estate sales by examnng the fnancng hypothess of Lang, Poulsen, and Stulz (1995). We explot the concept that nsttutonal nvestor nvolvement and debt oblgatons lead to effectve montorng of managers, compellng them to take value-maxmzng decsons and thus reducng the degree of agency costs of manageral dscreton. We show that the stock market responds more favorably to arm s-length corporate real estate sales by low agency-cost frm-years than those by hgh agency-cost frm-years. The result supports the fnancng hypothess that mples a negatve relaton between stock market responses to asset sales and degrees of agency costs. Keywords: Asset sales; Agency cost; Real estate; Corporate fnance; Tawan 1

I. Introducton Asset sales are common corporate actvty. How do these transactons affect shareholder wealth? Hte, Owers, and Rogers (1987, HOR hereafter) provde emprcal evdence that, on average, asset sales lead to sgnfcant share prce ncreases for sellng frms. Interpreted by HOR (1987), the postve prce reactons are because sellng frms capture some of the gans from allocatng assets to hgher-valued uses. In addton to supportng HOR s (1987) effcency deployment hypothess, John and Ofek (1995) also fnd evdence consstent wth ther own mprovng focus hypothess. In ther study, the announcement stock returns are greater for the sellers that ncrease focus n ther busness than others. In other words, asset sales elmnate negatve synerges between the sold asset and the remanng assets, thus ncrease share prces. Both the effcent deployment hypothess and the mprovng focus hypothess mply that shareholders beneft from asset sales equally whether management re-nvests or pays out the sale proceeds. Implctly the two hypotheses assume that management maxmzes shareholder wealth. 1,2 In the contrary, Lang, Poulsen, and Stulz (1995, LPS hereafter) advance and present emprcal support for ther fnancng explanaton of asset sales that arsng from management self-nterest pont of vew. Valung frm sze and control, management sells corporate assets to pursue ts own objectves that may ncrease or decrease shareholder wealth. Thus the frms, where agency costs of manageral dscreton are mportant, experence more favorable share prce reactons when payng out the proceeds than otherwse. In addton, consstent wth the fnancng hypothess, LPS (1995) show that the payout decson of sale proceeds s postvely correlated wth the manageral ownershp or nvestment opportuntes of the sellng frms. Bates 2

(2002) confrms the correlatons. Nevertheless, LPS (1995) do not fnd a drect lnk between stock-prce reactons and proxes for agency costs of manageral dscreton. On the other hand, Hrschey, Slovn, and Zama (1990, HSZ hereafter) have evdence that hgher stock returns n responses to general corporate dvestments made by U.S. frms are assocated wth hgher levels of bank debt. Lasfer, Sundarsanam, and Taffler (1996, LST hereafter) have the smlar fndng for market responses to general corporate sell-offs and levels of debt fnancng n U.K. The two studes both show that effcent montorng s postvely assocated wth market responses to asset sales and thus, provde more drect evdence lnkng stock-prce reactons and agency costs of manageral dscreton. The studes of general asset sales have been extended to focus on corporate real estate transactons. Examples are Glascock, Davdson, and Srmans (1991; GDS hereafter), Booth, Gloscock, and Sarkar (1996; BGS hereafter), and Lao and Chang (1996). However the studes have not yet examned LPS s (1995) fnancng hypothess. Ths study contrbutes to the lterature on corporate real estate sales by examnng the fnancng hypothess. Specfcally we examne the lnk between stock-prce reactons and agency costs of manageral dscreton n the context of corporate real estate transactons. Stratfyng observatons nto hgh agency-cost frm-years and low agency-cost frmyears wth montorng devces, ths study presents evdence that abnormal returns are hgher for low agency-cost frm-years than for hgh agency-cost frm-years n corporate real estate sales. Ths emprcal result supports the fnancng hypothess that agency costs of manageral dscreton matter n corporate real estate transactons. 3

The rest of the paper s organzed as follows. Secton II ntroduces the emprcal methods. Secton III descrbes the data. Secton IV presents emprcal results. The fnal secton offers some concludng remarks. II. Emprcal Methods To examne whether low agency-cost frm-years have hgher abnormal returns than hgh agency-cost frm-years n real estate sales, we use the market model as the returns generatng process 3 : R = α + β R + ε (1) t mt t where R t s the return on securty on day t, R mt s the return on the market ndex, Tawan Stock Exchange Captalzaton Weghted Stock Index (TAIEX), on day t, and ε t s a random error term. For each securty, the market model s computed over days 120 to 31 relatve to the event date of the sale. The 90-day perod conforms to the studes on asset sales of Ln and Shen (1996) n Tawan. We defne the dates when transacton contracts are sgned as the event dates. Abnormal returns for securty on day t are calculated as: AR = R αˆ + βˆ R (2) ( ) t t mt where the coeffcents α ˆ and β ˆ are ordnary least square estmates of the market model parameters for securty. Abnormal returns are then summed up over the perod from day T 1 to day T 2 ( T 1 < T 2 ) to obtan ndvdual cumulatve abnormal return defned as: 1 2 T2 CAR ( T, T ) = AR (3) t= T1 t day T 2 s: The cumulatve average abnormal returns over a sample of N securtes from day T 1 to 4

CAR( T, T ) = 1 2 N = 1 CAR ( T, T ) N 1 2 (4) The varance of CAR( T1, T2) s defned as: N N Var( CAR ( T1, T2)) 1 CAR ( T1, T2) Var( CAR( T1, T2)) = = CAR ( T1, T2) N N( N 1) = 1 = 1 N (5) Day T 1 to day T 2 s evaluated over the nterval from day 4 to day +2 relatve to the event date. The nterval concdes wth the market response pattern for real estate sales documented by Lao and Chang (1996). 4 The effect of asset sales on shareholder returns are tested for ts sgnfcance usng the t- statstcs defned as: t CAR CAR = (6) Var( CAR) Followng Noronha, Shome, and Morgan (1996), Lee, Pace, and Slawson (2002) and Lee and Slawson (2004), we stratfy frm-years nto a low agency-cost group and a hgh agencycost group wth montorng devces. Specfcally we group frm-years wth nsttutonal shareholdngs and book debt-to-asset debt ratos. Exstng studes generally support nsttutonal nvestor nvolvements and debt-fnancng serve as effectve montorng mechansm. Examples are Agrawal and Mandelker (1990), LST (1996), and Zantout (1997). In partcular, L, Wu, and Yeh (2003) and Kao, Chou, and Chen (2004) show nsttutonal nvestors and credtors are effectve montors n Tawan. We group the frm-years wth both nsttutonal ownershp and book debt-to-asset ratos below ther ndvdual 50 percentles nto the hgh agency-cost group and the other frm-years nto the low agency-cost group. 5 5

We test the dfference n cumulatve average abnormal returns between the hgh agency cost frm-years, CAR H, and the low agency-cot frm-years, CAR L, wth the statstcs defned as: t H L = CAR H Var( CAR ) + Var( CAR ) H CAR L L (7) where Var( CAR H ) and Var( CAR L) are the varances of the cumulatve average abnormal returns for the hgh agency-cost frms and low agency-cost frms respectvely. One concern s that the dfference between CAR and CAR reflects the removal of H L fnancal dffcultes f frms facng fnancal dstress domnate the sample of low agency-cost frms. LST (1996) present evdence showng that the beneft from asset sales comes from the resoluton of fnancal dffcultes. To address ths concern, we regress the cumulatve abnormal returns on degrees of agency costs and fnancal stuatons along wth other control varables: CAR = α+ β AG + β AG DC + β S + β S DC + β RG 1 2 3 4 5 + β FID + β ITD + β MRD + µ 6 7 8 (8) where AG measures the degree of agency costs, equals 1 for hgh agency-cost frm-years, and 0 for low agency-cost frm-years. DC s the dummy for fnancal dffcultes and equals 1 when the frm has a coverage rato n the year before the real estate sale below the sample medan and 0 otherwse. S s the relatve dvestment sze represented by the rato of the real-estate-sale prce to frm s equty market value at the 5 days mmedately pror to the event date. 6

RG s the relatve dvestment proft represented by the rato of the real-estate-sale proft (the dfference between the sellng prce and the apprased value) to frm s equty market value at the 5 days mmedately pror to the event date. 6 FID s the fnancng dummy equal to 1 when the statement of transacton purposes reveals that the sale s a fnancng actvty and 0 for other purposes. ITD s the nsder-tradng dummy equal to 1 when the drector/management net buys the corporate shares for the 6-month perod pror to the real estate sale and 0 otherwse. MRD s the market condton dummy equal to 1 when the sale occurred n a bull market and 0 n a bear market. µ s the error term. Equaton (8) s essentally an extenson of LST (1996). Nevertheless, we adopt the agency-cost dummy. In addton, followng LPS (1995), we nclude DC as the measure for fnancal dffcultes. We further nclude varables that are deemed to have nfluence on market reactons to asset sales documented n the lterature. Clubb and Stourats (2002) show that sell-off proftablty has mpacts on the market reactons to dvestture announcements by U.K. frms. To consder the effect, we ncorporate RG n the equaton. Followng Ln and Shen (1996), we tell asset sales for fnancng purposes from other sales accordng to the corporate statements of transacton purposes. The dummy FID ntends to capture the dfferental announcement effects of an asset sale whose purpose s fnancng other than other purposes. Ln and Shen (1996) use the fnancng dummy to capture agency costs proposed n LPS (1995). HSZ (1990) show that 7

nsder-tradng actvty conveys mportant nformaton to fnancal market partcpants about the valuaton of an asset sale event. Thus we ncorporate ITD nto our equaton. Lao and Chang (1996) ndcate that the market condton may affect market reactons to real estate transactons. We nclude MRD n equaton (8) to consder ths potental effect. Followng Lao and Chang (1996), we classfy our study perod nto bull markets and bear market by the overall trend of TAIEX. As expected, the Asa fnancal crss perod s classfed as a bear market. III. The Data We ntally purchase custom-bult data of fxed-asset sales of non-fnancal frms occurred durng March 7, 1994 to November 20, 2003 from Tawan Economc Journal (TEJ). 7 TEJ s the predomnant fnancal data provder for Tawan companes. Supplemented wth the M.O.P.S. of Tawan Stock Exchange and the udndata.com, we dentfy and focus our analyss on the subset of the TEJ data set that meets each of the followng crtera. Frst the events are real estate sales. We exclude the sales nvolvng non-real propertes. Ths ensures that the asset nature does not nfluence the return behavor around the events. Second, we exclude not-arm s-length sales. Not-arm s-length sales are where buyers who have stakes and/or potental nfluence on the corporate governance of sellers or vce versa. Ths excluson, smlar to BGS (1996), helps to focus on sales actually transferrng control. Thrd, we elmnate contamnated events. Specfcally we exclude sales that have earnngs or dvdends announced, or other real estate sales occurred n the twenty days surroundng ther event dates. We also elmnate one observaton potentally contamnated by a real estate related rumor. However we merge real estate sales by a frm n the same day. And fourth, samples wthout the other nformaton need to mplement the nvestgaton stated n the prevous secton are further elmnated. We obtan the 8

nformaton need to mplement the nvestgaton stated n the prevous secton from the usual TEJ databases and the M.O.P.S. Ths entre process reveals 81 real estate sales occurred durng the sample perod. Table 1 contans the descrptve statstcs of the fnal sample. Panel A shows the dstrbuton of the real estate sales over tme. The numbers of real estate sales start to ncrease dramatcally n 1997, reach ts peak n 1998, then declne and ease to the pre-1997 level from year 2001. The pattern may be caused by our sample selecton crtera. Nevertheless the pattern appears to concde wth the percepton that the Asa fnancal crss may affect corporate real estate holdng decsons. Panel B presents other descrptve nformaton about the sample. The mean sellng frm s market captalzaton s NT$ 39,728 mllon (medan NT$ 9,690 mllon) and the mean sellng prce of real estate s NT$ 653 mllon (medan NT$ 267). 8 The mean proft, the dfference between the sellng prce and the apprased value, from the real estate sales s NT$ 18 mllon (medan NT$ 6 mllon). The mean rato of sellng prce over the market captalzaton of the sellng frm s 19.526% (medan 4.409%) and the mean rato of proft over the market captalzaton of the sellng frm s 0.166% (medan 0.036%). 9 IV. Emprcal Results The secton presents our emprcal results. Table 2 summarzes the CARs for the full sample, and separately for the two hgh agency-cost and low agency-cost subsamples. The full sample has a sgnfcant CAR of 2.368%. The evdence confrms the general outcome of the exstng studes that sellers acheve postve unexpected returns at real estate transactons. That s, ths result supports the research of GDS (1991), BGS (1996), and Lao and Chang (1996). 9

The results of the subsamples are consstent wth the fnancng hypothess that mples a negatve relaton between stock prce effects and degrees of agency costs n asset sales. Specfcally the hgh agency-cost subsample experences no sgnfcant unexpected returns n ther real estate sales. The CAR for ths subsample s 0.350%. On the other hand, the low agency-cost subsample acheves sgnfcant postve unexpected returns. Ths subsample experences a CAR of 3.319%. As expected, the dfference n the CARs between the two subsamples s statstcally sgnfcant. The overall results of the subsamples suggest that the postve unexpected returns experenced by real estate sellers are drven by the low agency cost subsample. Table 3 reports the coeffcent estmates for the regresson model of Equaton (8). Same to Table 2, Table 3 also supports the fnancng hypothess. The coeffcent on the agency-cost varable AG s negatve and sgnfcant n all specfcatons. Ths ndcates a negatve relaton between stock prce effects and degrees of agency costs n real estate sales. The results are consstent wth the concept that nsttutonal nvestor nvolvement and debt oblgatons lead to effectve montorng of managers, compellng them to take value-maxmzng decsons and thus reducng the degree of agency costs of manageral dscreton. The coeffcent on the nteractve term AGDC s not sgnfcant. Ths does not support LST s (1996) dea that equty nvestors value montorng more when the sellng frm s fnancally dstressed and the bankruptcy event s avoded. It s lkely that our sample frms are fnancal healthy relatve to LST s (1996) frms and face no mmedate bankruptcy threat. However our results are consstent wth LPS (1995) who mply that the relaton between abnormal returns and agency costs does not depend on the sellng frm s fnancal stuatons. 10

The rato of the real estate sale prce to the sellng frm s equty market value does not appear to have nfluence on market responses. The postve sgn of the coeffcent on S s the same to the fndngs n prevous studes on asset sales. However the nsgnfcant coeffcent on S s dfferent from the sgnfcant and postve coeffcent documented n prevous studes on asset sales. These studes nclude Klen (1986) and LPS (1995). Klen (1986) suspects that the postve coeffcent occurs because the transacton prce relays nformaton about the net present value of the nvestment. Clubb and Stourats (2002) cast doubt on ths explanaton and thnk that the relatve transacton sze s lkely a proxy for the motvaton to ncrease focus. 10 Nevertheless LPS (1995) argue that relatve transacton szes should be negatvely correlated to stock market unexpected returns, f sales prmarly convey nformaton about the sellng frm s fnancng requrements and hence the frm s fnancal stuaton. Apparently there s no consensus on the expectaton of the relaton between transacton szes and market responses to asset sales. The nsgnfcant coeffcent n our sample suggests that the forces mentoned n the prevous studes may offset each other. The negatve coeffcent on SDC s consstent wth the expectaton of LST s (1996) expectaton that the stock market response to the transacton sze should be greater for fnancal dstressed frms. Nevertheless the coeffcent s not sgnfcant and thus does not provde support to ther expectaton. Interestngly LST (1996) themselves are not able to produce supportng evdence, ether. In fact, they fnd that the relatve sze of the dvestment s sgnfcant for the healthy, but not for the dstressed frms n nfluencng market reactons. The results are contrary to ther expectaton. The relatve dvestment proft RG has a postve coeffcent. However the coeffcent s not sgnfcant. The result does not support Clubb and Stourats (2002) who show sell-off proftablty has mpacts on the market reactons to dvestture announcements by U.K. frms. A 11

possble explanaton s the small magntude of the relatve dvestment proft n our sample. Eghty percent of the sample has the relatve proft only between -0.157% and 0.843%. The coeffcent sgns for the fnancng dummy FID and nsder-tradng dummy ITD are consstent wth the results of Ln and Shen (1996) and HSZ (1990). Nevertheless nether of the two coeffcent estmates s sgnfcant. Gven the agency cost varable AG n the equaton, ther nsgnfcance s not surprsng. Dfferent from Lao and Chang (1996), we fnd no evdence that the stock market responds more favorably to real estate sales n bull markets than n bear markets. It s lkely because Lao and Chang (1996) do not consder the other factors we consder n ths study. V. Concluson The fnancng hypothess of LPS (1995) mples a negatve relaton between stock market responses to asset sales and degrees of agency costs. Interestngly LPS (1995) do not fnd a drect lnk between stock-prce reactons and proxes for agency costs of manageral dscreton. Although HSZ (1990) and LST (1996) show such a lnk n not-real-estate sales, the exstng studes on corporate real estate sales has not yet examned LPS s (1995) fnancng hypothess. Supportng LPS s (1995) hypothess, our study provdes new evdence that agency-costs of manageral dscreton matter n market responses to asset sales n the context of corporate real estate transactons. Thus ths study not only adds to the lterature on corporate sell-offs but also contrbutes to those on corporate real estate sales. We explore the nfluence of montorng mechansm on market responses to arm s-length sales of corporate real estate. Consstent wth the concept that nsttutonal nvestor nvolvement and debt oblgatons lead to effectve montorng of managers, compellng them to take value-maxmzng decsons and thus reducng the degree of agency costs of manageral dscreton, our evdence shows that abnormal returns 12

are hgher for low agency-cost frm-years than for hgh agency-cost frm-years n corporate real estate sales. 13

Endnotes 1 Lang, Poulsen, and Stulz (1995) explctly pont out the shareholder wealth maxmzaton assumpton n the effcent deployment hypothess. 2 The nformaton hypothess s another explanaton that mples management sells corporate assets to maxmze shareholder wealth. Hte, Owers and Rogers (1987) examne and reject ths hypothess n the U.S. context. In addton the nformaton hypothess should be less applcable n the countres where asset revaluaton s permtted such as Tawan. 3 Booth, Glascock, and Sarkar (1996) use a market model that permts unexpected returns to follow a generalzed autoregressve condtonal heteroscedastc (GARH) process n ther study on corporate sell-offs of real estate assets n US. Ln and Shen (1996) also use a market model wth a GARCH process n studyng general asset selloffs n Tawan. Both of the studes conclude that the tradtonal OLS approach provdes overall economc results essentally dentcal to the GARCH model. 4 The length of the event nterval may reflect the nature of real estate transactons and the prce change lmts set n the Tawan stock market. Frms n a real estate transacton usually undergo a perod of negotaton and need to seek advces from real estate apprasal frms before sgnng a transacton contract. Therefore the market may know the nformaton before the contract dates. 5 The two montorng devces certanly are not all montorng devces and the groupng rule may not be the best. Thus our groupng may blur the dstncton between actual hgh-agency frms and actual low-agency frms. Fortunately, as ponted out by Lee and Slawson (2004) and Lee, Pace and Slawson (2003), ths consequence should create bas aganst us to fnd the sgnfcant dfference n market responses between classfed hgh-agency frms and classfed low-agency frms. 6 The M.O.P.S. of Tawan Stock Exchange provdes the apprased values of real estate made by real estate apprasal frms for each sale. 7 Startng from March 7, 1994, Tawan Stock Exchange requred frms to report ther sales wthn two days of the transactons occurred. 8 NT$34 was about US$1 on November 20, 2003. 9 A frm usually fnances ts real estate assets wth mortgages. Therefore the sellng prce can excess the market captalzaton of the sellng frm s equty stocks. That s, the rato of sellng prce over the market captalzaton of the sellng frm can be greater than 100%. 10 Ths explanaton does not apply to our study, snce none of our sample changes ther operatng actvtes materally from sellng real estate assets. 14

References 1. Agrawal, Anup and Gershon N. Mandelker., Large Shareholders and the Montorng of Managers: the Case of Anttakover Chater Amendments, Journal of Fnancal and Quanttatve Analyss, 1990, 25, 143-161. 2. Bates, Thomas W., Asset sales, Investment Opportuntes, and the Use of Proceeds, 2002, Workng paper, Unversty f Delaware. 3. Booth, G. Geoffrey, John L. Gloscock, and Sall K. Sarkar, A Reexamnaton of Corporate Sell-Offs of Real Estate Assets, Journal of Real Estate Fnance and Economcs, 1996, 12, 195-202. 4. Clubb, Coln and Ars Stourats, The Sgnfcance of Sell-Off Proftablty n Explanng the Market Reacton to Dvestture Announcements, Journal of Bankng and Fnance, 2002, 26, 671-688. 5. Glascock, John L., Wallace N. Davdson III, and C.F. Srmans, The Gans from Corporate Selloffs: the Case of Real Estate Assets, AREUEA Journal, 1991, 19, 567-582. 6. Hrschey, Mark, Myron B. Slovn, and Jans K. Zama, Bank Debt, Insder Tradng, and the Return to Corporate Selloffes, Journal of Bankng and Fnance, 1990, 14, 85-98. 7. Hte, Galen L., James E. Owers, and Ronald C. Rogers, The Market for Interfrm Asset Sales: Partal Selloffs and Total Lqudatons, Journal of Fnancal Economcs 1987, 18, 229-252. 8. John, Kose and El Ofek, Asset Sales and Increase n Focus, Journal of Fnancal Economcs, 1995, 37, 105-126. 9. Kao, Lanfeng, Jeng-Ren Chou, and Anln Chen, The Agency Problems, Frm Performance and Montorng: the Evdence from Collateralzed Shares n Tawan, Corporate Governance: An Internatonal Revew, 2004, 12, 389-402. 10. Klen, Aprl, The Tmng and Substance of Dvestture Announcements: Indvdual, Smultaneous and Cumulatve Effects, Journal of Fnance, 1986, 41, 685-696. 11. Lang, Larry, Annette Poulsen, and Rene Stulz, Asset Sales, Frm Performance, and the Agency Costs of Manageral Dscreton, Journal of Fnancal Economcs, 1995, 37, 3-37. 12. Lasfer M. Amezane, Pulyur S. Sundarsanam, and Rchard J. Taffler, Fnancal Dstress, Asset Sales, and Lender Montorng, Fnancal Management, 1996, 25, 57-66. 13. Lee, Mng-Long and V. Carlos Slawson, Jr, Montorng and Dvdend Polces for REITs under Asymmetrc Informaton, Paper presented at PRRES 10 th annual meetng, Bangkok, 2004. 14. Lee, Mng-Long, R. Kelley Pace, and V. Carlos Slawson, Jr., Dvdend Polces for REITs: Sgnalng v.s. Agency costs, Paper presented at AREUEA annual meetng, Atlanta, 2002. 15. L, Chun-An, Chn-Shun Wu, and L-Yu Yeh, Ownershp Structure and Illegal Corporate Act the Case of Tawan Lsted Company, Revew of Securtes and Futures Markets, 2003, 14, 75-138. 15

16. Lao, Hsen-hsng and We-Hua Chang, On the Effect of Real Estate Related Informaton on Stock Prce: an Emprcal Investgaton, Revew of Securtes and Futures Market, 1996, 8, 69-87. 17. Ln, Joungyol and Chung-Hua Shen, The Valuaton Effects of Announcements of Asset Selloffs, Revew of Securtes and Futures Market, 1996, 8, 1-22. 18. Noronha, Gregory M., Dlp K. Shome, and George E. Morgan, The Montorng Ratonale for Dvdends and the Interacton of Captal Structure and Dvdend Decsons, Journal of Bankng and Fnance, 1996, 20, 439-454. 19. Zantout, Zaher Z., A Test of the Debt-Montorng Hypothess: the Case of Corporate R&D Expendtures, The Fnancal Revew, 1997, 32, 21-48. 16

Table 1: Descrptve statstcs of the sample of real estate dspostons Panel A: Dsposton actvty by year Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Dsposton 1 8 2 15 22 11 13 3 5 1 Panel B: Characterstcs for the fnal 81 real estate dspostons Frm market value (NT$ mllon) Real estate sellng prce (NT$ mllon) Proft on the dsposton (NT$ mllon) (%) Mean 39,728 653 18 19.526 0.166 Maxmum 287,800 6,491 957 483.990 53.622 75% Percentle 30,274 705 16 7.809 0.198 Medan 9,690 267 6 4.409 0.036 25% Percentle 4,319 183 0 1.090 0.001 Mnmum 108 102-129 0.042-24.296 Notes. S s the relatve dvestment sze represented by the rato of the real-estate-sale sze to frm s equty market value at the 5 days mmedately pror to the event 1. date. 2. RG s the relatve dvestment proft represented by the rato of the real-estate-sale proft (the dfference between the sellng prce and the apprased value) to frm s equty market value at the 5 days mmedately pror to the event date. S RG (%) Table 2: Cumulatve average abnormal returns of the full sample and the subsamples Full sample (N=81) Hgh agency-cost sample (N=21) Low agency-cost sample (N=60) Dfference Interval CAR (%) t CAR (%) CAR t CAR (%) CAR t CAR th L -4 to 2 2.368 1.956* -0.350-0.307 3.319 2.113** -1.889* Notes. * Sgnfcant at the 0.1 level; ** Sgnfcant at the 0.05 level. 17

Table 3: Regresson analyss of CAR on the agency-costs of manageral dscreton Specfcaton 1 2 3 2 Adjusted R 0.190 0.216 0.010 Coeffcent t value Coeffcent t value Coeffcent t value Intercept 0.618 0.38 1.100 0.918 3.319 2.130 AG -2.851-1.675* -2.941-1.759* -3.669-1.916* AGDC 1.969 0.773 2.092 0.789 S 0.238 1.237 0.240 1.203 SDC -0.165-0.844-0.172-0.848 RG 0.141 0.349 0.165 0.415 FID -0.888-0.422 ITD 0.105 0.056 MRD 1.502 0.721 Notes. 1. AG measures the degree of agency costs, equals 1 for hgh agency-cost frm-years, and 0 for low agency-cost frm-years. 2. DC s the dummy for fnancal dffcultes and equals 1 when the frm has a coverage rato n the year before the real estate sale below the sample medan and 0 otherwse. 3. S s the relatve dvestment sze represented by the rato of the real-estate-sale sze to frm s equty market value at the 5 days mmedately pror to the event date. 4. RG s the relatve dvestment proft represented by the rato of the real-estate-sale proft (the dfference between the sellng prce and the apprased value) to frm s equty market value at the 5 days mmedately pror to the event date. 5. FID s the fnancng dummy equal to 1 when the statement of transacton purposes reveals that the sale s a fnancng actvty and 0 for other purposes. 6. ITD s the nsder-tradng dummy equal to 1 when the drector/management net buys the corporate shares for the 6-month perod pror to the real estate sale and 0 otherwse. 7. MRD s the market condton dummy equal to 1 when the sale occurred n a bull market and 0 n a bear market. 8. The t -statstcs are based on Whte s correcton for heteroscedastcty. 9. * Sgnfcant at the 0.1 level. 18