Should Earnings Thresholds be used as Delisting Criteria in Stock Market?

Transcription

1 Should Earnngs Thresholds be used as Delstng Crtera n Stock Market? 1. Introducton How should an emergng securtes market be regulated? What securtes laws work n an underdeveloped securtes market? Could there be any unntended consequences f unnecessary rules are adopted? These are tough questons that Chnese stock market regulators have encountered snce they jumpstarted the stock market wthn Chna s central plannng economy durng early 1990 s. Examnng the effect of securtes laws on stock market development n 49 countres, La Porta et al. (2005) fnd lttle evdence that publc enforcement beneft stock markets. In ths paper, we study a partcular provson n Chnese securtes laws and regulatons (specal treatment and delstng of stocks for reportng consecutve accountng losses). Consstent wth La Porta et al. (2005), we dentfy occasons where publc enforcement does not beneft the development of the underdeveloped Chnese stock market. Chnese securtes laws and regulatons 1 mandate that f a lsted frm reports accountng losses (.e., negatve earnngs) n two consecutve years, ts stock wll be put under Specal Treatment status (ST). There are varous tradng and fnancal restrctons on ST stock. Its daly stock prce movement s restrcted to be no more than fve percent n ether drecton, and the company s sem-annual report must be audted, unlke other companes. Furthermore, an ST frm cannot rase addtonal captal from stock market. If the frm reports one more loss, t s suspended from tradng on the stock exchanges, after a fourth annual loss the stock wll be delsted. In secton 2, we survey the small but growng lterature on securtes delstng (e.g., Macey et al., 2007), and motvate ths study. We argue that the ST polcy n Chna s securtes regulatons have serous unntended consequences. It could drve healthy frms out of the stock markets for temporary accountng losses (we refer to ths consequence as loss of economc effcency n that both lsted frms and nvestors are better off wth these frms stayng lsted). Out of fear of beng specally treated, lsted frms may also engage n value-destroyng earnngs manpulaton to prevent accountng losses (we refer to ths consequence as ncentve dstorton n that the polcy nduces frms to manage earnngs to avod losses). Secton 3 and secton 4 provde some exploratory evdence on these two unntended consequences. Secton 5 concludes the paper. 2. Lterature Revew and Motvatons Chna re-opened ts stock market n early 1990 s after a four-decade hatus. The purpose of such a move was clear: to help State-Owned Enterprses (SOEs) rase captal that the state government could no longer provde. However, n a centrally planned economy, the regulaton of 1 In the Unted States, lstng and delstng decsons are normally left to the stock exchanges (Securtes Exchange Act, 1934). But n Chna, Corporate Law of 1993 mandates that Chna Securtes Regulatory Commttee (CSRC, the counterpart of U.S. SEC) has the authorty to set delstng rules. Furthermore, although the Shangha and Shenzhen stock exchanges promulgate the mplementaton detals, they need CSRC s fnal approval. 1

2 ths new stock market was a challenge. Wth nvestor protecton n mnd, the regulators adopted securtes laws and regulatons from developed economes and made some add-on s, ST polcy beng one of these add-on s. The Securtes Law of 1993 mandates that lsted companes reportng three consecutve annual losses be suspended from tradng on the exchanges, and f they do not return to profts n the fourth year, the frms wll be delsted. Ths delstng provson stll exts n the most recent verson of the Securtes Law of Begnnng from Aprl 22, 1998, the Shangha and Shenzhen stock exchanges mplemented the specal treatment polcy (ST polcy). Lsted frms that reported two consecutve annual losses would henceforth be put nto ST status 2. There are varous tradng and fnancal restrctons on ST stock. Daly stock prce movement s restrcted to be no more than fve percent n ether drecton. In addton, ST company s sem-annual report must be audted, unlke other companes. Furthermore, an ST frm cannot rase addtonal captal from stock market. If the frm reports one more loss, t s suspended from tradng on the stock exchanges 3, and for a fourth consecutve annual loss, the stock wll be delsted 4. Ths ST polcy s stll effectve today. Between 1998 and 2004, 250 stocks were specally treated by Shangha and Shenzhen stocks exchanges 5. Between 1999 and 2005, 28 stocks were delsted from the Shenzhen Stock Exchange, and between 2001 and 2005, 14 stocks were delsted from the Shangha Stock Exchange. The reason we say that the ST polcy s an add-on provson to what Chnese regulators learned from other developed markets s that such a polcy s rarely seen n other securtes markets. As Macey et al. (2007) pont out, delstng rules can be dvded nto two categores. In the frst category are those rules desgned to nsure that the exchange s relatonshp wth the lsted frms reman proftable, as t s costly for exchanges to contnue to lst frms whose tradng s sporadc. In the second category are those rules that protect the reputaton of the exchanges. Therefore, exchanges around the world normally set mnmum prce, mnmum market captalzaton, and/or mnmum asset/revenue crtera for frms to contnue beng lsted. Macey et al. (2007) show that volaton of the mnmum stock prce standard s the most common cause of nvoluntary delstng from stock exchanges. It s very rare to see delstng requrements based on accountng earnngs. The ratonale used to base ST and delstng decsons on accountng earnngs (for brevty, ST polcy hereafter) s that frms reportng consecutve losses are poor performers (n the long run), and t s necessary to restrct or delst them n order to protect nvestor nterest. The man thess of ths paper s that we do not agree wth ths argument. Below we propose that the earnngs-based regulaton has two unntended consequences. Frst, frms temporarly reportng losses are not necessarly poor performng frms, especally n the long run. Thus the ST polcy could drve healthy frms out of stock market, or at least dsrupt frms operatonal strateges and dvert 2 Frms could be specally treated for other reasons, for example, negatve equty or fnancal fraud. But reportng losses are the domnatng cause for specal treatment. 3 From July 9, 1999 to May 1, 2002, suspended stocks were desgnated as Partcular Transfer (PT), whch allowed these stocks to be traded on the exchanges on Frdays under restrctve condtons. 4 The lstng manuals of the Shangha and Shenzhen stock exchanges, January 1, 1998 verson. On March 6, 1998, CSRC ssued a drectve approvng the ST polcy. 5 There were 851 lsted stocks n 1998, and the number was 1373 n

3 valuable management effort to focus solely on attanng profts. Ths leads to a net loss of economc effcency as both lsted frms and nvestors are worse off. Second, out of the fear of beng specally treated, lsted companes engage n value-destroyng earnngs manpulatons to avod reportng losses. Thus, the ST polcy dstorts management ncentves and nduces earnngs manpulaton behavor that actually harms nvestor nterest. To begn, t s worthwhle to note that there are legtmate reasons that a frm may report consecutve losses. Frst, accountng standards tend to be conservatve, and recognze expenses and losses more quckly than revenues and gans, resultng n a hgher occurrence of losses than otherwse (Basu, 1997). Second, early stage frms naturally tend to report more losses snce they are stll heavly nvestng n the busness, and the benefts of nvestment have not yet materalzed. Thrd, a macro-economc recesson could easly send many frms nto losses. Therefore, reportng losses does not necessarly mply poor performance n the long-run, and the ST polcy could thus drve healthy frms out of stock market. For example, should the same polcy exst n the Unted States, companes such as Apple Computers, Ford, and General Motors would have been specally treated at least once n ther lstng hstory, Yahoo and Lucent would have been suspended from tradng, and Amazon.com would have been delsted n Many other companes would also have been penalzed. Not only could the ST polcy drve healthy frms out of stock market, but t s also questonable why regulators have to specally treat actually poor-performng frms. Note that the ST polcy does not provde nvestors wth any new nformaton regardng the lsted companes. It s based solely on publcly avalable hstorcal earnngs, nformaton that nvestors already possess and have already responded to. Should the consecutve losses pertan to an actual reducton of frm value, the reducton would have already occurred and nvestor losses realzed (as stock prce declne) pror to the desgnaton of ST by stock exchanges. Furthermore, even f these stocks reported consecutve losses and suffered prce declne n the past, t does not mean they are low-future-return stocks. As Warren Buffet vews t, a good busness s not always a good purchase, although t s a good place to look for 6. That s because the busness brght future could be fully prced already. Smlarly, a bad (poor performng) busness s not necessarly a bad purchase f ts dsmal future s already fully prced 7, let along that past accountng losses do not necessarly translate nto dsmal future accountng performance. From an nvestor protecton perspectve, whch was the regulator s ntent, the ST polcy does no good. In fact, wth ths polcy, the regulators are judgng the value of lsted frms on nvestors behalf. There s anecdotal evdence that ths s not sound publc polcy. Rosenberry (1959) argues that a securtes market regulator should not pass judgment upon the value of securtes. Rosenberry (1959) ctes two early examples of delstng after the Securtes and Exchange Commsson made judgment on the value of lsted companes. The two frms nvolved were Standard Gas and Electrc Companes (1944) and St. Lous Southwestern Ralway (1944). In both cases, the SEC judged the common shares as worthless, and the New York Stock Exchange, n concurrence, suspended tradng on both stocks. But when the two stocks were later restored tradng, they sold at hgh prces. Today, t s mplctly a gudng prncple that market 6 Ten Great Investors. 7 Otherwse, we would have a hard tme explanng why there s so much tradng on OTC and pnk sheet markets. 3

4 regulators do not judge the value of lsted securtes on nvestors behalf (Macey et al., 2007). To summarze the frst unntended consequence of the ST polcy, the ST polcy does not provde nvestors wth new value-relevant nformaton, and more serously, could drve healthy frms that suffer short-term earnngs pressure out of the stock market. We provde exploratory evdence to shed lght on ths proposton n secton 3. The ST polcy then, as argued above, does not serve the ntenton of ts creators, and could nstead potentally do more harm to the market by nducng rampant earnngs manpulaton n lsted frms. Lsted frms manage earnngs for varous purposes, such as manageral compensaton, debt covenants, equty offerng, and to meet earnngs thresholds. Lsted frms also manage earnngs n response to regulatory requrements. For example, Jones (1991) examnes earnngs management of U.S. frms amed at recevng mport relef subsdy from federal government. Yang (2006) studes the reactons of lsted frms facng the threat of exchange delstng because of mnmum stock prce requrements. Yang (2006) shows that when delstng becomes hghly possble, frms wll engage n earnngs management to mprove stock prce and fend off the delstng threat. Not only are accrued earnngs managed to nflate earnngs, but these frms also cut R&D expendtures, and sell assets at gans to boost earnngs and stock prces. In the Chnese context, Dng et al. (2007), after surveyng earler studes, conclude that there s strong evdence that lsted Chnese frms manpulate earnngs to dramatcally boost ther earnngs to meet regulatory requrements to do rghts offerng and to avod specal treatment (.e., loss avodance). In partcular, Jang and Wang (2003), Jan and Wong (2006) and Lu and Lu (2007) document that the earnngs manpulaton technques used by Chnese frms to avod losses not only nvolve the tradtonally used accruals, but also nvolve the more destructve real-transacton based actvtes. Out of despar, lsted frms routnely engage n related party transactons, assets sales/purchases, equty sales/purchases, and restructurng. These transactons are often aded by controllng shareholders (proppng), but equally often result n subsequent tunnelng by controllng shareholders. In totalty, the ST polcy-nduced rampage of earnngs manpulaton dsrupts market order, damages nvestor confdence, and hnders market development. We beleve that ths s the second unntended consequence of the ST polcy. In secton 4, we provde a smple test for ths proposton. More supportng evdences can be found n the papers cted above and references theren. Our study contrbutes to the growng law and fnance lterature that studes how laws and regulatons facltate or hnder fnancal market development. In partcular, ths study jons a few other recent studes that examne how securtes market regulatons could have unntended consequences. Ayra et al. (2005) ponts out that although Regulaton Far Dsclosure Act (Regulaton FD) of 2000 was meant to curb the practce of lsted companes dsclosng value-relevant nformaton to a select subset of analysts and nvestors, so as to level the playng feld for all nvestors and analysts, the Regulaton FD may have the unntended consequences of encouragng herdng behavor among analysts and forcng companes to wthhold nformaton. As a result, Regulaton FD may actually reduce the total nformaton avalable to nvestors, contrary to the regulators ntenton. 4

5 Ayra et al. (2005) cte several emprcal studes that support ther arguments. Ewert and Wagenhofer (2005) study whether tghtenng accountng standards would restrct earnngs management. Whle t s wdely beleved that tghtenng accountng standards reduces earnngs management, Ewert and Wagenhofer (2005) shows that t s not always the case. Under tghtened standards, managers may substtute accrual-based earnngs management wth real transacton-based earnngs management. In fact, under certan condtons, even accrual-based earnngs management could ncrease due to the resultng ncreased senstvty of stock prces to earnngs. Overall, the tghtenng of standards, whch s meant to reduce earnngs management, could actually ncrease the extent of earnngs management n frms, wth the benefts of regulatory tghtenng outweghed by ts costs. Our study also contrbutes to the small but growng lterature on the economcs of exchange lstng and delstng. In partcular, a number of studes have challenged the current delstng standards n the Unted State markets, and call for changes or remedes. For example, Harrs, Panchapagesan and Werner (2006) examne 1098 delsted stocks from NASDAQ between 1999 and They fnd that almost ffty percent of regulatory delstngs volated only non-core requrements such as mnmum bd prce, but the adverse mpact of delstng on the frms and shareholders nvolved s huge. The authors suggest that NASDAQ consder revsng ts lstng crtera. Macey et al. (2007) call for smlar actons. 3. Specal Treatment Polcy and Economc Effcency In ths secton, we address the frst unntended consequence of the ST polcy. That s, we provde analytcal as well as collaboratve evdence to argue that the ST polcy could unntentonally drve healthy frms out of stock market when the ST, suspenson of tradng, and delstng decsons are based on accountng earnngs. To llustrate the problem caused by the ST polcy, we frst present a smple model to quantfy the probablty of a frm reportng consecutve losses. The purpose of the model s to compute, for a frm wth a certan level of long-term earnngs power (proftablty) and a certan level of fluctuaton n ts earnngs seres, the probablty that two consecutve losses would occur n ths earnngs seres. A key pont made by ths paper s that frms wth healthy long-term proftablty, due to large but temporary fluctuatons n ts operatons and earnngs, could be mstakenly specally treated. Let us consder the followng model, whch descrbes the earnngs seres of a frm. y = µ + σε, (1) t t where yt represents earnngs of the th frm n the t th year, µ represents the long-term average earnngs of the frm. In emprcal test below, µ s the average of hstorcal annual earnngs for a currently lsted frm. ε t s the year-to-year random varaton wth zero mean. 5

6 Earnngs volatlty s characterzed by σ, whch below s proxed by the standard devaton of a frm s hstorcal annual earnngs. For the purpose of smplcty, we assume that ε t ndependently follows a standard normal dstrbuton. 8 Based on model (1), the probablty of a frm experencng two consecutve losses can be drectly computed as follows: 2 µ P( yt < 0, yt+ 1 < 0) = P( yt < 0) P( yt+ 1 < 0) =Φ σ (2) Equaton (2) gves a parsmonous expresson of the determnants of the probablty of loss occurrence. The lower the long-term proftablty ( µ ), the more lkely a frm wll report two consecutve losses; the hgher the volatlty of earnngs ( σ ), the more lkely a frm wll report two consecutve losses. Therefore, reportng two consecutve losses does not necessarly mean the frm s poor performng n the long-run. Rather, t could be because the frm operates n more volatle ndustres thus has hgh earnngs volatlty 9. / µ σ, whch s a combnaton of the frm s long-term proftablty (.e., µ ) and earnngs volatlty (.e., σ ), better summarzes the determnants of loss occurrence. Intutvely, / µ σ characterzes a frm s proft stablty, hence we refer t as the proftablty stablty ndex (PSI). Hgher PSI frms are less lkely to report losses, and lower PSI frms are more lkely to report losses. It s worthwhle notng here that, when Chnese regulators set the ST polcy, they reasoned that consecutve loss frms had lttle nvestment value, and to protect nvestors, these frms should be watched and restrcted, even delsted f more losses followed. It s hghly probable that the regulators beleved that these loss-generatng frms were low µ frms, and gnored the fact that hgh earnngs volatlty could easly send hgh µ frms nto consecutve losses n certan years. Armed wth ths smple statstcal model, we can emprcally test the probablty of healthy frms beng specally treated when they should not be. Ths exercse s necessary because f there s no economc consequence to the ST polcy, we mght not be nterested n removng t. 8 Note that n the real data, t s more realstc to assume a heavy-taled dstrbuton wth a tme-seres dependence structure for ε t. However, such an assumpton can only make the statstcal analyss more sophstcated and wll not nvaldate our conclusons. In fact, by assumng a heavy-taled dstrbuton wth a tme-seres dependence structure, the tal probablty as characterzed by model (2) wll be even larger, whch mples that our results as presented n Table 1 understate the probablty of specally treatng healthy frms. 9 Thnk about the ol companes whose earnngs are tghtly related to crude ol prce. In late 1990 s, when crude ol prce was around $10 per barrel, ol companes earnngs, f not n absolute losses, were not even comparable to what they would be today, when crude ol prce s close to $100 per barrel. 6

7 However, we should not use data from Chnese frms to evaluate ths model because Chnese lsted frms are already subject to the ST polcy, and may have already managed ther earnngs to avod losses (Dng et al., 2007; Jang and Wang, 2003; Jan and Wong, 2006; Lu and Lu, 2007). What we are tryng to test, on the other hand, s the probablty of losses n the absence of the ST polcy. If we use Chnese earnngs data, we would lkely underestmate the probablty of loss occurrence, thus underestmate the probablty of a (long-term) healthy frm reportng two consecutve annual losses. Put t dfferently, n Chnese market, fnancally-troubled frms would be mstakenly dentfed as healthy frms, because they have managed earnngs to report profts n order to get around the ST threat. As a result, we wll actually understate the consequences of the ST polcy. Therefore, we chose to use the U.S. data to evaluate our model. U.S. frms are not subject to the ST polcy, or other smlar polces, so earnngs should not have been managed to avod specal treatment. To the extent that U.S. frms also prefer small profts to small losses, and have managed earnngs to avod losses for ths reason (Burgstahler and Dchev, 1997), the results of our test mght also underestmate the probablty of a (long-term) healthy frm reportng two consecutve annual losses, smlarly causng an understatement of the consequences of the ST polcy (drvng healthy frms out of stock market). But ths bas should be much smaller than tests usng Chnese data because the loss avodance ncentve for U.S. frms are much smaller than that for the Chnese frms gven the ST polcy. The dataset we use s taken from the COMPUSTAT database and ncludes all frm-years from 1975 to 2003 whch have the necessary data tems. We get a total of 90,070 annual observatons coverng 8,392 stocks. On average, a total of 10 years observatons were collected for each frm. Based on these observatons, the values of µ and σ are estmated by usng sample mean and sample standard devaton of return on equty (ROE) for each frm durng ths sample perod. Then the value of PSI can be estmated for each frm. We sort all 8,392 stocks on ther PSI values, and compute the probablty of each frm reportng both one annual loss and two consecutve losses n ts hstory accordng to equaton (2). In Table 1, we report the PSI and the probablty of losses for those frms whose PSI s between the lower quartle and the medan n the cross secton. We chose to report ths group of frms because they are defntely healthy frms n that ther long-term proftablty s good. However, they have a relatvely hgh probablty of reportng two consecutve accountng losses as well due to relatvely hgh earnngs volatlty (.e., relatvely low proftablty stablty). Table 1 shows that the PSI values of the twenty-fve percent of all frms fall between and The assocated one annual loss probablty ranges from 42.7% to18.3%, whle the probablty of reportng two consecutve losses ranges from 18.2% to 3.3%. Were the ST polcy mplemented n the U.S. market, a frm whose PSI was better than twenty-fve percent of all frms would face the unreasonably hgh probablty of 18.2% to be specally treated. Even for the medan PSI frm, there stll exsts a 3.3% probablty that t would be specally treated Note that such an estmate s actually very conservatve due to the fact that ε t s assumed to be condtonally ndependent and normally dstrbuted. 7

8 To gauge how proftable these frms are, we pck two frms whose PSI are at 0.184, the 25 th percentle, and fnd ther hstorcal average ROE to be 10% and 13%, very healthy returns to nvestors, but wth a ST polcy, these two frms would face a 18.2% chance of beng specally treated. Gvoly and Hayn (2000) document that the proporton of loss frms n the U.S. COMPUSTAT unverse has been steadly and monotoncally ncreasng over the last few decades, from 1.67% durng the perod, to 15.48% durng the perod when NASDAQ frms were added, and to 33.58% durng the perod. Even a constant sample of 896 frms reports an ncrease n the proporton of loss years from 7.68% durng the perod to 20.29% durng the perod. The change n the constant sample ndcates that the ncreasng occurrence of losses s not purely due to addng young, hgh-tech frms, who tend to experence more losses, to the sample. However, the ncreasng occurrence of losses n lsted companes does not mean the lsted frms, or the stock market as a whole, are losng nvestment value. In fact, the contrary s true. Therefore, to summarze ths secton, wth the collaboratve support from a smple loss-generatng model, and the fact that losses have becomng more frequent n the Unted States market (whch does not mean that the U.S. market s losng nvestment value), we beleve the ST polcy n Chna s market has the unntended consequence of possbly drvng healthy frms out of the stock market, or at least dsruptng the normal operaton of lsted frms as well as dvertng valuable manageral effort to unnecessarly mantanng postve earnngs. All these amount to a loss of economc effcency. 4. Specal Treatment Polcy and Earnngs Manpulaton Whle the ST polcy drves healthy frms out of the stock market at an unacceptable level of probablty, t also nduces currently lsted frms to engage n value-destroyng actvtes n order to avod reportng a loss. That s, facng the possblty of ST and possbly delstng, company management has a stronger (dstorted) ncentve to avod losses, even f t means manpulatng earnngs and decevng nvestors. Dng et al. (2007), Jang and Wang (2003), Jan and Wong (2006), Lu and Lu (2007), among others, have carefully documented that Chnese lsted companes not only manage accruals, but also engage n real transacton-based technques to dramatcally boost earnngs to avod reportng losses. These real transactons nclude related party transactons, asset sales/purchases, equty sales/purchases, restructurng, or outrght fraudulent actvtes. In ths secton, we employ the methodology developed n Burgstahler and Dchev (1997) to provde supportve evdence of earnngs manpulaton to avod losses by Chnese frms. Ths methodology has already been used wdely to test earnngs management n varous contexts (Degeorge et al., 1999; Dechow et al., 2003; Xue, 2004). 11 In Chnese context, ths methodology has been used by Haw et al. (2005) and Chen et al. (2001), among others. Haw et al. (2005) observe that the percentage of Chnese lsted frms 11 Durtsch and Easton (2005) questoned the valdty of ths methodology; therefore, we need to nterpret results generated by ths methodology wth cauton. 8

9 reportng ROE between 10 and 11 percent durng perod s 23.8 per cent, whle t s only 8.98 percent durng the perod. A Chnese CSRC regulaton became effectve n 1996, whch requred that n order to qualfy for rghts offerng, a lsted frm must mantan ROE no less than 10 percent n the three years pror to rghts offerng. Haw et al. (2005) fnd evdence that managers execute transactons nvolvng below-the-lne tems and use ncome-ncreasng accruals to meet regulatory benchmarks for a rghts offerng. Chen et al. (2001) show a sgnfcantly postve assocaton between lsted frms recevng modfed audtor opnons (MAOs) and reportng profts margnally above the target levels specfed n stock de-lstng and rghts offerng regulatons, and nterpret these fndngs as consstent wth the noton that asymmetrc proftablty requrements exacerbate managers' propensty to engage n earnngs management whch n turn s postvely assocated wth recevng MAOs. Followng these pror studes, ths paper uses the same methodology. Namely, we compare the relatve densty of small proft frms vs. small loss frms n the annual cross-secton. In Fgure 1, we plot hstograms of annual return on equty (ROE) for both U.S. frms and Chnese frms 12. Each bar corresponds to a 2% ROE nterval. The dotted lne s the zero earnngs lne. To the rght of t are postve earnngs, and to the left of t are losses. We defne frm-years whose ROE les between 2% and 0% as small loss frm-years, and those between 0% and 2% as small proft frm-years. A jump of ROE densty across pont zero (zero earnngs) can be easly dentfed n both U.S. and Chnese frms. In the U.S., frms also prefer profts to losses, and tend to manpulate earnngs to a certan extent to turn a small loss year nto a small proft year (Burgstahler and Dchev, 1997). Fgure 1 shows that the densty of small proft frm-years s 0.69% hgher than the densty of small loss frm-years n the U.S. sample. However, t s also clear from Fgure 1 that the ncentve to avod loss s much stronger for Chnese frms than for U.S. frms. The densty of small proft frm-years s 7.05% hgher than the densty of small loss frm-years n the Chnese sample. The jump of densty from small loss to small proft for Chnese frms s nearly ten tmes as large as that for the U.S. frms. Furthermore, Chnese frms are only half as lkely as the U.S. frms to report losses, small or large. The strong nclnaton for Chnese frms to avod loss s an ndcaton that ST polcy dstorts management ncentves and nduces pervasve earnngs manpulaton n the Chnese market. To drectly lnk the tendency of loss avodance n Chnese frms to the enactment of ST polcy, we check whether the proporton of small loss (small proft) frms has decreased (ncreased) wth the ntroducton of ST polcy. Specfcally, for each year, we count the number of frms whose reported ROE fall nto the ntervals of [-10%, 0%] and [0%, 10%], respectvely. 13 We denote the correspondng counts by 0 0 n and. Furthermore, we use n + n and n + to denote the total number of frms wth negatve ROE (.e., loss frms) and postve ROE (.e., proft frms), respectvely. Usng ths, the magntude 12 Chnese stock data comes from CCER SINOFIN database. Sample for Fgure 1 s taken from 1998 to Usng [-5%, 0%] and [0%, 5%] to measure small loss and small proft does not qualtatvely change our results. However, usng further narrower ntervals s not feasble because annual observatons are too few n some years to do meanngful comparson. 9

10 of the tendency of loss avodance (earnng management) s quantfed as 0 0 n n + Percentage Dfference = 100% n+ n (3) Ths percentage dfference measures the dfference between the proporton of small proft frms n all proft frms and the proporton of small loss frms n all loss frms. The results are plotted n Fgure 2. ST polcy was frst ntroduced nto the market n As Fgure 2 shows, before 1998, the percentage of small loss frms n all loss frms s slghtly larger than the percentage of small proft frms n all proft frms. But snce ST polcy was nstalled, the percentage of small proft frms has steadly ncreased relatve to the percentage of small loss frms, by an ever-wdenng margn. In 2002, the dfference was reached more than 40%. If anythng, Fgure 2 shows the altered behavor of management n reacton to the nstallaton of ST polcy. Under the ST polcy, lsted companes are more averse to reportng losses than wthout the polcy n place, and they manpulate ther earnngs n order to report profts 14. Takng Fgure 1 and Fgure 2 together, n conjuncton wth the emprcal results from Dng et al. (2007), Jang and Wang (2003), Jan and Wong (2006), Lu and Lu (2007), and others, we beleve that ST polcy has nduced Chnese lsted companes to engage n earnngs manpulaton to avod losses. As a result, not only were nvestors suppled wth msleadng fnancal nformaton, but also valuable management tme and effort has been dverted from managng the frms and wasted. 5. Concludng Remarks The purpose of ths paper s smple: to document a stuaton where over-regulaton n an underdeveloped stock market has hndered ts development. The ST polcy n Chnese stock market causes economc effcency losses (drvng healthy companes out of the stock market), and dstorts management ncentves hereby nducng rampant earnngs manpulaton. After the enactment of two stock exchanges n 1990 and 1991, Chnese stock market flourshed durng most of the 1990 s, as a new channel of fnancng n socalst Chna. But after the stock market peaked n 2000, t langushed durng the next fve years. Came along wth the market stagnaton were revelatons of outrageous manpulatons of earnngs, wdespread tunnelng by controllng shareholders (Jang et al., 2005), and large-scale nsder tradng. Measures taken by the central government to revve the stock market largely faled to boost nvestor confdence n the stock market, and the market was n the verge of beng margnalzed n the fast-growng Chnese economy. In 2005, Chnese regulators started a reform to make non-tradable state and nsttutonal shares tradable. The reform has largely been completed by Ths reform also sparked a fast 14 Chnese economy had been growng steadly durng our sample perod, so another nterpretaton for Fgure 2 s that the proftablty of lsted Chnese frms grew wth the overall economy. However, we carefully documented the proftablty of lsted frms durng our sample perod, and t actually declned almost monotoncally. Ths makes our argument that the ncreasng frequency of small proft frms result from loss avodance even stronger. The proftablty data s avalable upon request. 10

11 and phenomenal stock prce boom. However, at ths moment, whether ths boom s sustanable or not s yet to be seen. Furthermore, the ST polcy s stll n effect today. If the nternet bubble and the subsequent revelaton of accountng scandals of Enron, WorldCom and the lke tell us anythng, t s that lsted companes, hopng to sustan hgh stock prces, have even stronger ncentve to manpulate earnngs n a bull market. Specal treatment s the last thng lsted frms need. Fnally, whle we recognze the exploratory nature of our study, we beleve that more research on the economcs of lstng and delstng s warranted, not only n Chna, but also n developed securtes markets. Today, the technology for securtes tradng and nformaton dssemnaton and ntegraton s fundamentally more advanced than when most of today s lstng and delstng rules were formulated, and thus changes are necessary for these rules not to hnder market development (Macey et al., 2007). In Chna s case, whle the government allowed stock markets to be establshed n 1990, t has tred consstently to keep the market under tght control ever snce, partcularly n the lstng and delstng areas. Ths study calls for changes n the ST polcy, but other areas, such as daly prce lmt and short-sale constrant, need also to be examned. 11

12 References Arya, A., Glover, J., Mttendorf, B., Narayanamoorthy, G., Unntended consequences of regulatng dsclosures: the cases of regulaton far dsclosure. Journal of Accountng and Publc Polcy 24, Basu, S., The conservatsm prncple and the asymmetrc tmelness of earnngs. Journal of Accountng and Economcs 24, Burgstahler, D., Dchev, I., Earnngs management to avod earnngs decreases and losses. Journal of Accountng and Economcs 24, Chen, C., Chen, S., Su, X., Proftablty regulaton, earnngs management, and modfed audt opnons: evdence from Chna. Audtng: A Journal of Practce and Theory 20, Dechow, P., Rchardson, S., Tuna, I., Why are earnngs knky? an examnaton of the earnngs management explanaton. Revew of Accountng Studes 8, Degeorge, F., Patel, J., Zeckhauser, R., Earnngs management to exceed thresholds. Journal of Busness 72, Durtsch, C., Easton, P., Earnngs management? the shapes of the frequency dstrbutons of earnngs metrcs are not evdence pso facto. Journal of Accountng Research 43, Dng, Y., Zhang, H., Zhang, J., Prvate vs state ownershp and earnngs management: evdence from Chnese lsted companes. Corporate Governance 15, Ewert, R., Wagenhofer, A., Economc effects of tghtenng accountng standards to restrct earnngs management. Accountng Revew 80, Gvoly, D., Hayn, C., The changng tme-seres propertes of earnngs, cash flows and accruals: has fnancal reportng become more conservatve? Journal of Accountng and Economcs 29, Harrs, J., Panchapagesan, V., Werner, I., Out but not gone, a study of Nasdaq delstngs. Workng Paper, Unversty of Delaware. Haw, I., Q, D., Wu, D., Wu, W., Market consequences of earnngs management n response to securty regulatons n Chna. Contemporary Accountng Research 22, Jan, M., Wong, T. J., Proppng and tunnelng through related party transactons. Revew of Accountng Studes. forthcomng. Jang, G., Lee, C., Yue, H., Tunnelng n Chna: the pervasve use of corporate loans to extract funds from Chnese lsted frms. Workng Paper, Cornell Unversty and Pekng Unversty. Jang, G., Wang, L., How far would management go to manage earnngs? Workng Paper, Pekng Unversty. Jones, J., Earnngs management durng mport relef nvestgatons. Journal of Accountng Research 29, La Porta, R., Lopez-de-Slanes, F., Shlefer, A., What works n securtes laws? Journal of 12

13 Fnance 61, Lu, Q., Lu, Z., Corporate governance and earnngs management n the Chnese lsted companes: a tunnelng perspectve. Journal of Corporate Fnance, forthcomng. Macey, J., O Hara, M., Pomplo, D., Down and out n the stock market: the law and fnance of the delstng process. Journal of Law and Economcs, forthcomng. Rosenberry, S.L., Lstng and delstng securtes on the New York Stock Exchange. Vrgna Law Revew 45, Xue, Y., Informaton content of earnngs management: evdence from managng earnngs to exceed thresholds. Workng Paper, MIT. Yang Y., Earnngs management as a response to the threat of exchange delstng. Workng Paper, Unversty of Colorado, Boulder. 13

14 Table 1: The Estmated Proftablty Stablty Index and Assocated Probablty of Frms Reportng Annual Losses PSI Percentle PSI Probablty of Reportng One Annual Loss Probablty of Reportng Two Consecutve Annual Losses 50% % 3.3% 45% % 5.5% 40% % 7.9% 35% % 10.6% 30% % 13.6% 25% % 18.2% Note: Ths table reports the Proftablty Stablty Index (PSI) and the probablty of annual losses for those frms whose PSI s between the lower quartle and the medan n the cross secton of 8,392 U.S. frms. For each frm, PSI equals to / µ σ, where µ s the average of the tmes seres of reported annual ROE of the frm n the COMPUSTAT fle, andσ s the standard devaton of the annual ROE n ths tme seres. The probablty of reportng one loss, condtonal on / gven by ( 0) ( 0) µ σ, s µ µ P yt < = P µ + σεt < = P εt < = Φ. The probablty of σ σ reportng two consecutve losses, condtonal on µ / σ, s gven by 2 µ P( yt < 0, yt+ 1 < 0) = P( yt < 0) P( yt+ 1 < 0) =Φ σ. 14

15 Fgure 1: Hstograms of Return on Equty (ROE) of U.S. and Chnese Frms 15

16 Fgure 2: Percentage Dfference between Proporton of Small Proft and Proporton of Small Loss Frms n Chnese Stock Market by Year Note: Each year, we compute the number of small proft frms (ROE between 0% and 10%) as a percentage of all proft frms, and the number of small loss frms (ROE between 10% and 0%) as a percentage of all loss frms. Percentage dfference s the dfference between the two percentages. 16