The real effects of Chinese stocks delisting : evidence from a difference in difference model

(1)

The Real Effects of Chinese Stocks Delisting:

Evidence From a Difference in Difference Model

Master thesis

Version: Final

06 Jul 2015

Student Name: Pengting Dai

Student Number :10825975

Supervisor: dr. R. (Rafael) Almeida da Matta

MSc

Business Economics

(2)

2

Statement of Originality

This document is written by Student [Pengting Dai] who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

2011 is a disaster year for China Stocks in US exchanges. A rash of accounting scandals of China stocks come to public attention, more than 100 China firms listed on the NASDAQ were delisted or suspended during 2011 and 2012. My interesting originated from this mass delisting event of China stocks. Compared to IPO, delisting got fewer attentions in literature. Past literature consistently show market quality and liquidity deteriorates significantly around the delisting event itself. base on exiting theories, I examined real effects of delisting on delisted China stocks during 2011 and 2012 on four dimensions, daily price, market liquidity, trading volume and daily return. Further, I checked the effect of delisting on four dimensions in immediate term, short term, and long term separately. Collected data of 34 China delisting firms in treatment group and 177 China listing firms in control group, from Wharton Research Data Services, Google. Finance and Website of firms. Use Difference in Difference model to analysis the real effect on four dimensions and add additional control variables to do robustness checks. Finally, study results reveal price of the delisting China firms suffer average 81% decreasing after delisting in immediate term and suffer 83% decreasing in long term. Market liquidity of delisting firms is significantly destroyed. Effective spreads increase approximately 400% in immediately term, 128% in short term and 122% in long term. Trading volume fluctuate largely one month before delisting, which represents react of investors to delisting risk. Trading volume declines by 73% in short term and 80% in long term. Return decreases significantly only in immediately term, while increases significantly

in long term.

Key words: Delisting Difference in Difference Daily price Market liquidity Trading

volume Daily return Trend

(4)

4

1. Introduction

Since Chinese companies begun to list on the NASDAQ, NYSE and AMEX in United States in 1990. IPO oversea became more and more popular for Chinese firms which eager to seek for financial capitals. Besides the common advantages for a firm go to listing, IPO in US exchange have more special advantages for Chinese firms. Firstly, list on foreign market seems more difficult than list on domestic market. Actually, in some ways, listing overseas is easier, cheaper and fairer. Most firms sacrifices rather more than we can imagine or saw in surface to successful IPO in China. Compared to large invisible cost, reverse merger is a relatively easy and quick way for Chinese firms go IPO overseas. Secondly, listing on US exchanges creates a liquid market outside of domestic country for Chinese shareholders. Thirdly, firms’ reputation can largely improve by listing on US exchange. More attractive to customers and suppliers, even easier to get credit from bank. So China firms rush into US exchange. During 2007 and 2010, 159 Chinese companies entered the U.S. securities markets using reverse mergers and generated market capitalization of $12.8 billion. In the same period, even 56 Chinese companies had an aggregate market capitalization of

$27.2 billion1. However, 2011 year came with big disaster for Chinese public

companies listed in the U.S. Specifically; dozens of Chinese firms were delisted from major U.S exchanges because of accounting irregularities. In report of Sarah Mish kin (2012), the value of Chinese companies delisting from US exchanges in 2011 exceeded the value of money Chinese companies raised via US initial public offerings. Even no firm IPO successfully in quarter 4, 2011. Compared to 2010, Chinese companies raised only $2.2billion through IPOs in 2011, only half of amount they raised in 2010. This scandal made many foreign investors bearish on Chinese listing firms. Even, a large amount of China firms still listing and operate well on US

1

(6)

2

exchange were unfortunately called junk stocks.2

My interesting originated from this mass delisting event in 2011. The effects of listing shares on an organized exchange have been studied extensively. Compared with listing study, delisting has received less attention in the literature. The existing literature consistently show market quality and liquidity deteriorates significantly around the delisting event itself. Abase on existing literatures, I planned to study on real effect of 2011 China delisting scandal on four dimension, daily price, trading volume, market liquidity and daily return of delisting firms. Further, studied the real effect of delisting on four dimensions of delisting stocks in different sub-periods.

Learn from event study. I create three windows [-5/+60], [-60/+60], [-450/+450], as proxies for immediate term, short term an long term respectively. Selected Difference in Difference model to study the real effect of delisting, with eight

hypotheses. My sample period is January 3rd 2011 to December 28th 2012. Control

group sample are 177 China firms listing on NASDAQ in sample period. And treatment group sample are 34 China delisting firms which were delisted in sample period and were continually trading on OTC market. I collected trading data and financial fundamental data from CRPS, CompUSA and IBES of Wharton research data service, Google. Finance and website of firms. Finally, only reject Hypothesis 4.2. The results reveal that average daily price of delisting firm significantly decrease 81% in immediate-term and 83% in long-term though delisting. Effective spreads increase significantly near four times in immediately-term, increase 104% in short-term, and increase 87% in long-term. One month before delisting, the trading volume had abnormally increasing and larger fluctuating. Trading volume significantly declines 73% in short-term and 61% in long-term. Daily return decrease significantly in immediately-term; Also not significantly decreasing in short term; And increase significantly in s long-term.

2

(7)

3

2. Background and Literature review

2.1 Background

2.1.1 Delisting in US exchange

The removal of a listed security from the exchange, on which it traded before, is called delisting. A firm may be involuntary delisted by the SEC for rule violations, or voluntary delisting for mergers and acquisitions. There are complex requirements of corporate governance, market capitalization, price, net asset, number of shareholder, number of market maker and so on for firms which want to continually listing on NASDAQ stocks. In involuntary delisting, any firm fails to meet the minimum standards would trigger the delisting process. In the delisting process, the firm would get a deficiency notice firstly. The notice informs the firm the way it can survival from involuntary delisting in 90 calendar days. If the firm still failed to survival by the way deficiency notice told, it would receive a "determination letter" of NASDAQ, asking the firm delisting. However, facing delisting determination of NASDAQ, the firm has the right to appeal to NASDAQ's listing qualifications panel; or choose compliance possess. If both were failed, it would be delisted finally. In United States, companies delisted from the NASDAQ can continue to trade on the over-the-counter markets or the Pink Sheets. Even though, delisting largely degrade firm’s credit among lenders. What’ more, ruin the firm’s reputation. So in existing literature, market quality and liquidity deteriorates significantly during delisting.

2.1.2 China stocks Delisting scandal

In 2011, a rash of fraud and accounting scandals of China stocks came to public; more than 100 Chinese companies listed on the NYSE were delisted or suspended from trading through 2011 and 2012. The plunge in share prices as cross-border Chinese

(8)

4

listings collapsed wiped out more than $40 billion of value, sending investor sentiment into a deep freeze and ushering in an era of mistrust of US-traded Chinese

shares.3 Also, the listing value decreasing largely. There were 41 firms IPO in 2010,

11 firms IPO in 2011, only 2 firms IPO successfully in 2012. In report of Sarah Mishkin

(2012)4, the value of Chinese companies delisting from US exchanges in 2011

exceeded the value of money Chinese companies raised via US initial public offerings. Even no firm IPO successfully in quarter 4, 2011. Compared to 2010, Chinese companies raised only $2.2billion through IPOs in 2011, only half of amount they raised in 2010 . This scandal made many foreign investors bearish on Chinese listing firms. Even, a large amount of China firms still listing and operate well on US

exchange were unfortunately called junk stocks.5

2.2 Literature review

Since 2004, delisting rate increased approximately as much as 20% annually. Nearly half of these firms are forced to delist by NASDAQ and most subsequently trade via the over-the-counter (OTC) Bulletin Board or the Pink Sheets. Active post-delisting makes it possible to study on the changes of different dimensions through delisting event. However, the lack of market data on delisted firms makes delisting studies relative difficult.

Existing literatures consistently shows market quality and liquidity deteriorates significantly around the delisting event itself. However, those literatures only give some general conclusion with delisting. Specifically, Sanger and Peterson (1990) study the involuntary delisting from New York Stock Exchange (NYSE) and America Stock Exchange (ASE), and doucument a fall of about 8.5% in the stock price of

3

A easy ‘ 26 Chinese Firms Delisted in 2011’. A website ‘ on the website of Business Insider UK’

4_{Stanley Lubman, (2013). What U.S.-China Auditing Dispute Means for Chinese Business Culture.}

China Real Time

5

(9)

5

delisted firms on the delisting announcement day. Shumway (1997) demonstrates an average delisting return of −30% for the firms that are delisted for bankruptcy and other negative reasons., Harris, Panchapagesan and Werner (2008) study the involuntary delisting from NASDAQ, and report that investors experience a loss of about 22% in 60 days prior to delisting. Share volume declines by two-thirds, quoted spreads increase from 12.1% to 33.9%, and effective spreads triple from 3.3% to 9.9%. Park and Lee (2014) shows that the spread triples and the volatility doubles in the OTC market after involuntary delisting, reducing in liquidity as well as causing the fall of stock prices due to delisting. Further, different samples and measurement methodologies from difference view always give different results. Sun, Tang and Tong (2002) studied the mass delisting Malaysia Stocks from exchange of Singapore (SES) from 1989 to 1998 by OLS model with dummy variables. The mass delisting increase Malaysia home betas and decreasing Singaporean market betas. Also, reduces the total liquidity of both markets. Park, Lee and W. Park (2014) study involuntary delisting in Korea exchange from January 2003 to December 2012, using buy-and-hold abnormal returns in event study. Firms delisted by the Korea show a price decline long before the delisting decision is made. The mean buy-and-hold abnormal returns for one year prior to the delisting decision are −53.30% for the Main Board sample and −75.88% for the KOSDAQ firms, indicating that most of the delisted firms face financial distress prior to delisting. The lack of liquidity for delisted stocks appears to account for the huge drop in price in the Korean market. In this sense, all those result is consistent with the liquidity hypothesis of Macey et al. (2008) and Sanger and Peterson (1990), that reduction in liquidity is the primary reason for decreasing of stock price.

Delisted firms and listing firms have some difference before delisting. Bakke, Jens and Whited (2012) shows the delisted firms are much smaller, are much more highly levered, have much more negative earnings. Difference in difference model can alleviate those differences. Further, each firm has its own specific characteristics.

(10)

6

Control variables are very important in explaining the firm specific characteristics, eliminating endogenous problem. Beavera, McNicholsa and Priceb(2007) controlled firms characters by earning ,cash flow and book-to-market. Park, Lee and W.Park (2014) used marketing size m growth rate, standard deviation of return, turnover rate as control variable to specify the firms’ characters. In Regression Discontinuity Design model of Bakke, Jen and whited (2012), control firms by sales, earning, and market-to-book. Bae and Wang (2012) study on the Chinese listed stock performance in US , Firm Characteristics by Market capitalization, Debt to assets and Cash.

2.3 Hypothesis and contributions

My paper focus on the China stocks delisting from US exchange in 2011 and 2012. There are no literature systematically study on the China delisting stocks in this special year . Further, I not only study on price, liquidity, trading volume and return changes after delisting, but also effect of delisting in difference sub-periods. I study those changes for immediately term, short term and long term separately. Specifically, I have eight hypotheses. Hypothesis 1.1 immediate-term effects and long-term effects of delisting on daily price of delisting stocks are significantly different from zero. Hypothesis 1.2 only immediate-term effects of delisting on trend of delisting stocks daily price is significantly different from zero. Hypothesis 2.1 immediate-term, short-term and long-term delisting effects on delisting stocks market liquidity are all significantly different from zero. Hypothesis 2.2 immediate-term, short-term and long-term delisting effects on trend of delisting stocks market liquidity are all significantly different from zero. Hypothesis 3.1 short-term and long-term delisting effect on delisting stocks trading volume are significantly different from zero. Hypothesis 3.2 only short-term delisting effect on trend of delisting stocks trading volume is significant different from zero. Hypothesis 4.1 immediate-term, short-term and long-term delisting effects on delisting stocks daily return are all significantly different from zero. Hypothesis 4.2 immediate-term, short-term and long-term

(11)

7

delisting effects on trend of delisting stocks daily return are all significantly different from zero. In all, I use Difference in Difference methodology research on the real effect of delisting on four dimensions in different periods. My new angle of view on delisting effect would contribute to the limited number of delisting studies.

3. Methodology

3.1 Sample

3.1.1 Treatment sample

Firms in treatment group should meet three criterions. Firstly, firms were IPO before 01Jan2011; Secondly, firms were involuntarily delisted from January 2011 to December 2012; thirdly, firms should continually trade on the OTC Bulletin Board (OTCBB) or in the Pink Sheets after delisting. Fourth, the data of firms is accessible. Then collect daily price, ask and high, bid and low, trading volume, share outstanding , daily return , earning, sales, debt to assets, cash, debt to assets, book to market of each stock in trading days or fiscal quarter.

3.1.2 Control sample

Firms in treatment group should meet three criterions. Firstly, firms were IPO before January 2011; Secondly, firms were not delisted before December 2012, if firms were delisting one day; fourth, the data of firms is accessible. Then collect daily price, ask and high, bid and low, trading volume, share outstanding , daily return , earning, sales, debt to assets, cash, debt to assets, book to market of each stock in trading days or fiscal quarter.

(12)

8

3.2 Econometrics model

3.2.1 Difference in Difference model

Difference in Difference model can eliminate the change of stocks in control group and get the pure change of stocks in treatment group. Further, to solve the exogenous problem in regression. I check the standard controls in previous literatures. Bae and Wang (2012) study on the performance of China stocks in US exchanges. They using the control variables: Cash, market capitalization and Debt to assets to control specific characteristics of China firms listed on US exchanges. Both of us research on the China firms listed on US exchanges, so I use Market capitalization, Debt to assets and Cash as well, and adding volatility of return as Park, Lee and W.Park (2014) do.

3.2.1.1 Value change in DID of four dimensions

Four dimensions Value change can get from regression

2

1 2 3 4

* /

it i it it it it it return it it

Y   treat post treat post spread cash maketsize D A   _ u

DID effect Firms specific characters: standard control

Cont

2 1cashit 2maketsize 3D A/ it 4 return it

     _

rol variables Treatment group

Treatment: Event window [-360/360],[-60/+60],[-60,+5]

Control group

Treatment Effect Change in control group

Change in Treatment group

(13)

9

, 0

, , , , 0

Y Price

Y Spread Effectivespread Tradingvolum Return 



 

 

Con : Mean outcome of control group before time of treatment

Treat: Difference of mean outcome: treatment group vs control group before

delisting

Post: Difference of mean outcome: Control group after time of delisting relative to

control group before

Treat*post: Additional change in value of treatment group (relative to control

group) around treatment.

3.2.1.2 Adding Trend change in DID of five dimensions

Five dimensions trend change can get from:

1 2 3

* *

* * /

it i it it t i t it i

t it i it it it

Y treat post treatpost dTrend eTreat Trend fPost Treat gTrend Post Treat cash maketsize D A

                    , 0 , , , , 0 Y Price

Y Spread Effectivespread Tradingvolum Return 



 

 

Trend d : Trend of control firms before treatment.

Treat*trend e : Difference in trends between treated vs control firms before treatment

Post* trend f : Change in trend in control firms around treatment

Treat*post*trend g : Additional change in trend of treaded firms (relative to control firms) around treatment.

(14)

10 3.2.1.3 Sample period and Sub-period

Sample period are from January 2011 to December 2012. To test the change in different terms, I choose three different windows. Specifically, use [-60/+5] as a proxy for immediately reaction. [-60/+60] as a proxy for short term. [-450/+450] as a proxy for long term. For each of four dimensions (Price, Market liquidity, Trading volume, Return), run three times regression in the three sub-period [-60/+5], [-60/+60], [-450/+450] respectively.

3.2.2 DID in four dimensions

3.2.2.1 Daily price

Hypothesis 1.1 immediate-term effects and long-term effects of delisting on daily

price of delisting stocks are significantly different from zero. Hypothesis 1.2 only immediate-term effects of delisting on trend of delisting stocks daily price is significantly different from zero.

1 2 3

* *

* * /

it i it it it i t it i

t it i it it it

P treat post treatpost spread eTreat Trend fPost Trend gTrend Post Treat cash maketsize D A

    

   

       

   

Do above regressions respectively in [-450/+450], [-60/+60], [-60/+5]. Firstly, I expected that ___{450/ 450}_ _ ,___{60/ 5}__ are negative and are significantly different from zero, which means price value of delisted firms are significantly decreasing in immediate term and in long term. If results meet my expectation, Hypothesis 1.1 cannot be rejected. Secondly, I expected g___{60/ 60}_ _and g___{450/ 450}_ _ am not significantly from zero, which means price trend of delisted firms are not significantly different from price trend before delisting. While g___{60/ 5}__is significantly different from zero, which means price trend changes significantly only in immediate term. If the result is in my expectation, Hypothesis 1.2 cannot be rejected.

(15)

11 3.2.2.2 Market liquidity

Most of the post-delisting price becomes as low as less than 0.5, so the absolutely spread is not comparative to the spread before delisting. Here I use effective spread instead of spread r to measure the market liquidity.

Hypothesis 2.1 immediate-term, short-term and long-term delisting effects on

delisting stocks market liquidity are all significantly different from zero. Hypothesis

2.2 immediate-term, short-term and long-term delisting effects on trend of delisting

stocks market liquidity are all significantly different from zero.

1 2 3

* *

* * /

t it i it it it

eS treat post treatpost spread eTreat Trend fPost Trend gTrend Post Treat cash maketsize D A

    

   

       

   

Do above regressions respectively in [-450/+450], [-60/+60], [-60/+5]. Firstly, I expected to get ___{450/ 450}_ _,___{60/ 60}_ _,___{60/ 5}__ all are Positive and are significantly different from zero ,which means market liquidity of delisted firms are significantly decreasing in immediate term, short term and long term. If the results are in my expectation, Hypothesis 2.1 cannot be rejected. Secondly, I expected

 60/ 5

g_ _ g___{60/ 60}_ _and g___{450/ 450}_ _ are all significantly different from zero, which means market liquidity trend of delisted firms are significantly different from the trend of listing firms in short term and long term. If the result is in my expectation,

Hypothesis 2.2 cannot be rejected. 3.2.2.3 Trading volume

Hypothesis 3.1 short-term and long-term delisting effects on delisting stocks trading

volume are significantly different from zero. Hypothesis 3.2 only short-term delisting effect on trend of delisting stocks trading volume is significant different from zero.

(16)

12 1 2 3 * * * * / it i it it it i t it i t it i it it it

V treat post treatpost spread eTreat Trend fPost Trend gTrend Post Treat cash maketsize D A

    

   

       

   

Do above regressions respectively in [-450/+450], [-60/+60], [-60/+5]. Firstly, I expected to get ___{450/ 450}_ _,___{60/ 60}_ _are negative and are significantly different from zero ,which means trading volume of delisted firms are significantly different from volume in pre-delisting, in short term and long term. If the results are in my

expectation, Hypothesis 3.1 cannot be rejected. Secondly, only g___{60/ 60}_ _ is

significantly different from zero, which means trend of post-delisting trading volume is significant different from the trend of pre-delisting trading volume in short terms. . If the results are in my expectation, Hypothesis 3.2 cannot be rejected.

3.2.2.4 Daily return

Hypothesis 4.1 immediate-term, short-term and long-term delisting effects on

delisting stocks daily return are all significantly different from zero. Hypothesis 4.2 immediate-term, short-term and long-term delisting effects on trend of delisting stocks daily return are all significantly different from zero.

1 2 3

* *

* * /

t it i it it it

R treat post treatpost spread eTreat Trend fPost Trend gTrend Post Treat cash maketsize D A

    

   

       

   

Do above regressions respectively in [-450/+450], [-60/+60], [-60/+5]. Firstly, I expected to get___{450/ 450}_ _ ,___{60/ 60}_ _ ,___{60/ 5}__ all are significantly different from zero ,which means post-delisting return of delisting stock is significantly different from pre-delisting return in three terms. Also, expect ___{60/ 5}__is negative and___{60/ 60}_ _,

 450/ 450

_ _ are positive, which means the return decrease in immediate term but

(17)

13

Hypothesis4.1 cannot reject. Secondly, I expected g___{60/ 5}__ g___{60/ 60}_ _and g___{450/ 450}_ _

are all significantly different from zero, which means daily return trend of delisted firms are significantly different from the trend of listing firms in immediate term, short term and long term. If the results are in my expectation, Hypothesis 4.2 cannot be rejected.

4.Date and descriptive statistics

4.1 Data

All the stocks in control group and treatment group are stocks of China firms, trading on US NASDAQ exchange before delisting.

4.1.1 Treatment group

Filter treatment group by the four criteria. Firstly, firms were IPO before 01Jan2011; Secondly, firms were involuntarily delisted during January 2011 to December 2012; thirdly, firms should continually trade on the OTC Bulletin Board (OTCBB) or in the Pink Sheets after delisting. Fourth, the data of firms is accessible. There are 51 delisting stocks I can find of during sample period, 8 stocks with Delisting code 200~300, delisting for ’merger and acquisitions’ while 43 stocks with Delisting code 500~600,delisting for ’drop’. only 34 stocks were continually trading over-the-counter are accessible. So those 34 delisting firms are in treatment group. Specifically, pre-delisting data and post-delisting data of 34 stocks come from CRPS database of Wharton Research Data Services, and Google. Finance or website of firms respectively. Also, the post-delisting data in Google.Finance are incomplete, so I search the data from internet, especially website of the firm as supplement.

(18)

14

daily) and IBES database (Unadjusted Details/ Detail History Actuals, Unadjusted) during sample period from January 1st 2011 to December 31th 2012 ,collect variables : Price, High or ask, Low or bid, Holding period return, Number of shares outstanding, Trading volume, Delisting Cod, Date of Delisting Payment, Company Legal Name, Current ISO Country Code Headquarters, Earning per share, Book value, Market value, Sales, Debt, Total assets and Cash. Then, I collect the post-delisting data from Google. Finance and website of firms. Open price, high price, low price, close price, trading volume are easy to accessible. Also, found the data of earning per share, market value, book value, sales, cash, debt, totally asset in financial statements of each firm manually. Besides, I instead missing data of yearly average value. Then merge the pre-delisting dataset and post-delisting dataset together. Create three different event windows [-60/+5], [-60/+60], [-450/+450] from the panel dataset. In [-60/+5] sub-period, there are 2244 observations in treatment group. In [-60/+60] sub-period, there are 4114 observations in treatment group. In [-450/+450] sub-period, there are 30634 observations in treatment group.

4.1.2 Ccontrol group

From CRSP, Compustat and IBSE, collect Price, High or ask, Low or bid, Holding period return, Number of shares outstanding, Trading volume, Delisting Cod, Date of Delisting Payment, Company Legal Name, Current ISO Country Code Headquarters, Earning per share, Book value, Market value, Sales, Debt, Total assets and Cash. Drop the incomplete date. Then 177 China stocks left as the Control group. Compared to treatment group, it is easier to get the date of Chinese firms in control group. All the trading data and financial statement data are provided Wharton Research Data Services database. Also, Create three different event windows [-60/+5], [-60/+60], [-450/+450] from the panel dataset. In [-60/+5] sub-period, there are 11682 observations in control group. In [-60/+60] sub-period, there are 21417 observations in control group. In [-450/+450] sub-period, there are 159477 observations in control group.

(19)

15

4.2 Statistic summary

As in the graphs shows, all the abnormal dramatically changes happen in short term windows [-60/+60].

Figure 1,2,3,4

Average Trend of Four Dimensions in [-60/+60] window

177 China stocks listing on US NASDAQ duiring period 2011 to 2012 are control group and 34 China stocks delisting from NASDQ, then continully to trading on OTC market duiring period 2011 to 2012 are treatment group.. there are 34 different delisting data. Each stock creat 34 event window [-60/+60]. Event time equals 0 is the delisting data. For treatment group there are 1156 [-60/+60] event windows,each window have 121 observations . For control group there are 6018 [-60/+60] event windows, each window have 121 observations.For each stock in each day has one price value, effextive spread value, trading volume value and return value.

Figure1. Average Daily Price of Two Samples In Short Term. A point in red line on x

day is acculated as the mean of whole 1156 daily prices on the same x day in 1156 event windows. A point in black line on x day is acculated as the mean of whole 6018 daily prices on the same x day in 6018 event windows.

(20)

16

Figure2. Average Effective Bid-ask Spread of Two Samples In Short Term. A point in

red line on x day is acculated as the mean of whole 1156 effective bid-ask spreads on the same x day in 1156 event windows. A point in black line on x day is acculated as the mean of whole 6018 effective bid-ask spreads on the same x day in 6018 event windows.

Figure3. Average Trading Volume of Two Samples In Short Term. A point in red line

on x day is acculated as the mean of whole 1156 trading volumes on the same x day in 1156 event windows. A point in black line on x day is acculated as the mean of whole 6018 trading volumes on the same x day in 6018 event windows.

(21)

17

Figure4. Average Daily Return of Two Samples In Short Term. A point in red line on

x day is acculated as the mean of whole 1156 daily return on the same x day in 1156 event windows. A point in black line on x day is acculated as the mean of whole 6018 daily return on the same x day in 6018 event windows.

4.2.1 Daily price

Figure 1 shows there is no great change on price of control group. Specifically, the daily stock price of listing firms, reported in panel A of Table 1, varies between 1.7 and 142.83 around mean price 11.91 before delisting. The daily price of them in panel B varies from 0.93 to 140.63 around mean price 10.31 after delisting. Also, Figure 1 shows a dramatically decreasing of delisting firms during delisting date of treatment group. Specifically, the daily stock price of delisting firms in short term, reported in panel A of table 1, varies from 0.17 to 14.62 around mean price 4.2 before delisting. The daily price of them in panel B varies between 0.03 and 4.1 around mean price 0.52 after delisting.

(22)

18 Ta b le 1 D escr ip tiv e Sta tis ti cs o n F o u r Di m e n si o n s o f Co n tr o l G ro u p an d T re atm e n t G ro u p 17 7 Chin a st o cks lis tin g o n U S N A SD A Q d u irin g p eriod 2011 t o 2012 ar e co n tr o l gr o u p an d 34 Chin a st o ck s d eli stin g fr o m N A SDQ , th en c o n tin u lly to tr ad in g o n O TC m ar ke t d u irin g p eriod 2011 t o 20 12 a re t re at m en t gr o u p .. th er e a re 34 d if fe re n t d eli stin g d at a. E ach s to ck cre at 34 e ve n t w in d o w [ -60/ +6 0] . E ve n t ti m e eq u al s 0 is t h e d elis tin g d at a. I re gar d d eli stin g d at e in p o st -d eli stin g p eriod . Pre -d e lis tin g p e riod is [ -6 0,0), w h ile p o st -d eli stin g p eriod i s [0,+60 ]. For tr ea tm en t gr o u p th er e ar e 1156 [ -60 /+ 6 0] e ve n t w in d o w s, each w in d o w h av e 1 21 o b se rv atio n s . Fo r co n tr o l gr o u p th er e ar e 6018 [ -60/ +6 0] e ve n t w in d o w s, e ach w in d o w h av e 121 o b se rv at ion s.F o r each s to ck in e ach d ay h as o n e p ri ce v alu e, e ff ext iv e s p re ad v alu e , tr ad in g vo lu m e v alu e an d r etu rn v alu e. Colu m (A ) is th e n u m b er o f o b se rv atio n s. Fo r co n tr o l gr o u p (p an el A 3 6108 0 an d p an l B 36 709 8 ), n u m b er o f o b se rv at ion s eq u als 7 2817 (60 1 8 m u tip ly 121) ; Fo r tr e at m en t gr o u p ( p an el A 6936 0 an d p an l 70516 ) ,e q u als 13 98 76 . Colu m e (B ) is th e n u m b er o f firm s in gr o u p . Colu m (C ) is t h e m ean o f var iab le o f each o b se rv atio n s. Colu m (D) i s th e s tan d ar d d ev ia tio n o f var iab le in s u b -s am p le s. Colu m n (K ),(J ) ar e re sp e ctiv ely th e m in im u m v alu e a n d ma xi m u m v alu e in sub -s amp le s. Tr ad in g v o lu m e a re r ep re se n t in M illi o n (M) . Da ily r etu rn a re p re se n t in p er ce n at ag e ( % ). Wi n d o w [ -60 /+ 60] C o n tr o l G ro u p Tr e atment Gr o u p (A ) (B) (C) (D) (E ) (F ) (G ) (H) (I) (J) (K) (L ) V ar iab le O b s N Me an Std . De v. Mi n Max O b s N Me an Std . De v. Mi n Max Pa n el A . Pre -d el is ti n g p er io d Price 361080 177 11.91 17.38 1 .7 0 142.83 69360 34 4.20 3.19 0.17 14.62 eS p re ad 361080 177 0.05 0.05 0.00 0.7 8 69360 34 0.09 0.08 0.00 1.15 Trad in g V o lu me ( M) 361080 177 0.48 1.38 0.00 38.7 0 69360 34 0.33 0.82 0.00 16.00 R etu rn ( %) 361080 177 -0.35 4.22 -42.13 78.26 69360 34 -0.13 12.94 -10 0 .00 51 .21 Pa n el B . Po st -d el is ti n g p eri o d Price 367098 177 10.31 17.11 0 .93 14 0 .6 3 70516 34 0.52 0.48 0.03 4.10 eS p re ad 3 67098 177 0.06 0.05 0.00 3.50 70516 34 0.20 0.24 0.00 3.52 Trad in g V o lu me ( M) 367098 177 0.48 1.45 0.00 20.80 70516 34 0.09 0.33 0.00 6.60 R etu rn ( %) 367098 177 -0.30 5.2 7 -10 0 .0 0 149.53 70516 34 -5.49 45.55 -10 0.00 62.50

(23)

19 Ta ble 2 A ve ra ge pe rce n t c ha n ge in f o ur dim e n sion s be tw e en p re -de lis ti n g a nd po st - de list ing The nu m be r in each cel l repr es ent t he per cent ag e cha ng e o f var ia bl es fr o m p re -del is tin g to po st -del is tin g. P er cent ag e is ca lc ul at ed by the ‘g ap bet we en po st -del is tin g valu e an d pre -del is tin g valu e’ to pre -del is tin g valu e. E.g . The val ue in ( C) o f da ily pric e -1 3 .4 2 % i s der ivat iv e fr o m di ff er enc e o f two m ean pric es (1 0 .3 1 , 1 1 .9 1 ) in t ab le 1 co lum n (C) o f da ily pric e. Di vi de by pr e -d eli st in g pric e 1 1 .91 . Tab le pr esent t he pe rc ent ag e ch an ge in m ean an d s tan da rd deviat io n o f var ia bl es i n t hre e di ff er ent t er m s. C o n tr o l G ro u p Tre at m en t Gro up (A ) (B ) (C) (D) (E ) (F ) (G ) (H) (I) (J) (K) (L ) Sub - p eri o d [-5/+ 60] [-60/+ 60] [-450/+ 450] [-5/+ 60] [-60/+ 60] [-450/+ 450] V ar iab le s Me an Std . D Me an Std . D Me an Std . D Me an Std . D Me an Std . D Me an Std . D Dai ly P ri ce -3.45 2.1 7 -13.42 -1.55 -5.84 -2.18 -84.94 -7 7.29 -8 7.6 2 -85.12 -89.65 -8 7.04 Eff ec ti ve S p re ad 22.01 24.25 24.44 20.36 35.99 31.05 416.58 568.82 128.0 7 20 2.7 1 122.46 160.56 Trad in g vo lu me 0.91 -0.58 0.00 5.34 0.6 7 6.56 30.95 18.91 -73.26 -60.1 7 -79.30 -6 7.55 R etu rn ( %) 45.48 17.0 2 -14.4 7 24.7 2 35.32 38.25 35903.7 6 1014.91 4149.68 251.91 -281.75 5.89

(24)

20

(continued in Page 17)

Further, Volatility of daily price is measured by the standard deviation of the price. In treatment group, the standard deviations of daily price has a decrease from 3.19 to 0.48, totally decrease 85.12% itself, reported in Table 2 column (J), in short term. At the meantime, in control group, the standard deviations of daily price has a slightly -1.55%, reported in Table 2 column (D), decrease from 17.38 to 17.11 in short term.

4.2.2 Liquidity : Effective Ask-bid spread

Figure 2 shows no great change in effective spread of control group through delisting. Specifically, the effective ask-bid spread listing firms, reported in panel A of Table 1, varies between 0 and 0.78 around mean price 0.05 before delisting. The effective ask-bid spread of them in panel B, varies from 0 to 3.50 around mean price 0.06 after delisting. In all, reported in column B of Table 2, the effective ask-bid spread increasing 24.44% through delisting. It is small change, compared with 128.07% increase in effective spread f delisting stocks, reported in in column (I) of Table 2, in short term.

Also, Figure 2 shows there is a dramatically increasing in effective ask-bid spread of treatment group through delisting. Specifically, the effective ask-bid spread listing firms, reported in panel A of table 1, varies from 0 to 1.15 around mean price 0.09 before delisting. The effective ask-bid spread of them in panel B, varies from 0 to 3.52 around mean price 0.2 after delisting. In all, reported in column E of table2, the effective ask-bid spread increasing absolutely 129%, relatively 104% (eliminates 24.44% of control group) through delisting.

Further, reported in Table2, the average percentage change of listing firms in three terms are 24.25% in immediate term, 20.36% in short term and 31.05% in long term. While the percentage change in delisting firms are much larger 416.58% in

(25)

21

immediate term, 128.07% in short term and 122.46 % in long term. As the terms become longer, the effective spread trend difference between control group and treatment group become smaller.

4.2.3 Trading volume

Figure 3 shows no great change in trading volume of control group through delisting. Specifically, the trading volume of listing firms, reported in panel A of table 1, varies between 0 to 38.7 million around mean value near 0.48 million before delisting in short term. The trading value of them in panel B varies from 0 to 20.8 around the same mean value near 0.48 million after delisting in short term. In all, reported in column B of table2, the trading volume of listing firms have approximately zero change through delisting in short term.

Figure 3 shows there is an abnormal volatility of trading volume one month before delisting, while there is a dramatically decreasing in trading volume after delisting. Specifically, the trading volume of delisting firms, reported in panel A of table 1, varies between 0 to 16 million around 0.33 million before delisting. The trading volume of them in panel B, varies between 0 to 6.6 million around the 0.09 million. In all, reported in column B of table2, the trading volume of delisting firms have approximately 73 % decreasing through delisting.

Reported in table2, the average percentage changes of standard deviation of delisting firms are -18.91% in immediate term, 60.17% in short term and 67.55% in long term. The percentage change in immediate term is obviously different from in the other two terms. Present in figure3, because of the abnormal change in 30days before delisting, rather than the change after delisting.

(26)

22

4.2.4 Daily Return

Figure 4 shows no great change in daily return of control group through delisting. Specifically, the daily return of listing firms, reported in panel A of table 1, varies from -42.13% to 78.26% around mean return -0.35% before delisting. The daily return of them in panel B, varies from-100% to 149.5% around the mean return -0.3% after delisting. In all, reported in column B of table2, the daily return of listing firms decreasing 14.47 % through delisting in short term.

The daily return of delisting firms, reported in panel A of table 1, varies from -100% to 51.21% around -0.13% before delisting. The daily return of them in panel B, varies from-100% to 62.5% around the-5.49%. In all, reported in column B of table2, the daily return of listing firms have approximately 4149.68% decreasing through delisting in short term.

5.Result

My results are in line with the general theory existing in previous literatures. In general, delisting decrease the price value, destroy the market liquidity, decrease trading volume and increase return. However, there are some difference in sub-periods. My study provides more detail analysis in different periods on 2011 China accounting scandal background in US exchange.

5.1.Daily price

Regress by Difference in difference model in three different sub-periods, with the standard control: cash, market capitalization and debt to assets ( Bae and Wang 2012). In my study, Hypothesis 1.1 immediate-term effects and long-term effect.

(27)

23 Ta b le 4 . De lis ting e ff e ct o n f o u r d im e n si o n s in th re e t e rm s Dai ly P ri ce (A ) ( B) (C) Ef fe cti ve S p re ad (D ) (E ) (F ) Tr ad in g V o lu me (G ) (H ) (I ) Dai ly R etu rn (J) (K ) (L ) [-5/+ 60] [-60/+ 60] [-450/+ 450] [-5/+ 60] [-60/+ 60] [-450/+ 450] [-5/+ 60] [-60/+ 60 ] [-450/+ 450] [-5/+ 60] [-60/+ 60] [-450/+ 450] Tr ea t -5.304 *** -5.440 *** -4.20 7 *** 0.0396 *** 0.0393 *** 0.0105 0.140 * 0.118 * -0.0551 2.534 *** -0.3 75 -8.8 70 *** (-10.4 7) (-10.9 7) (-4.09) (13.7 8) (11.31) (1.60) (2.4 7) (2.14) (-0.46) (4.12) (-0.64) (-12.99) Po st -0.811 -0.7 42 ** -0.542 0.010 7 * 0.003 78 * 0.00 26 2 -0.0 253 -0.0 240 0.0184 0.661 0.210 -0.89 7 ** (-0.93) (-3.14) (-1.1 7) (2.15) (2.28) (0.89) (-0.26) (-0.9 2) (0.34) (0.63) (0.7 6) (-2.91) Tr ea t* p o st -3.1 78* * * -1 .358 -3.314* ** 0.265 *** 0.125 *** 0.0833 *** 0 .258 -0. 25 4 ** -0. 238* * -75.14 *** -0.593 7.468 *** (-4 .21 ) (-1 .0 9) (-3. 78 ) (19.4 7) (24.7 0) (8.64) (0.98) (-3.04 ) (-3.01 ) (-26.25) (-0.69 ) (7.41) Tr ea t* Tre n d -0.019 7 -0.0195 0.0149 0.000 221 ** 0.000 22 7 * -0.000 1 63 0.00618 *** 0.0059 2 *** 0.00 269 -0.0 738 *** -0. 108 *** -0.256 *** (-1.40) (-1.41) (0.61) (2.7 7) (2.35) (-1 .32) (4.01) (3.89) (0.95) (-4.35) (-6.66) (-15.7 4) Po st* Tr e n d -0.232 -0.033 7 *** -0.0 4 57 *** -0.000160 0.000 249 *** 0.000329 *** -0.00 295 0.000110 -0.00130 0.148 0.0 234 *** 0.0 215 ** (-0.89) (-5.5 7) (-4 .28) (-0.11) (5.90) (4.83) (-0.10) (0.16) (-1.05) (0.4 7) (3.31) (3.0 2) Tr ea t* Po st* Tr end 0.220 *** 0.0645 0.0308 -0.01 72 *** -0.00181 *** 0.000509 * -0.0886 -0.00666 ** -0.00125 24.35 *** 0.29 2 *** 0.264 *** (3.7 7 ) (1 .14) (0.85) (-4.40) (-12.56) (2.21) (-1.18) (-2.94) (-0.30) (30.23) (12.12) (10.9 7) Sp re ad 16.95 *** 16.32 *** 16.51 *** (119.05) (16 2.7 0) (132.60) St an d . c o n tr o l x x x x x x x x x x x x Co n tr o l_ co n s 4.434 *** 4.80 7 *** 4.988 *** 0.05 21 *** 0.05 28 *** 0.0518 *** 0.0948 *** 0.11 7 *** 0.1 47 *** 4.896 *** 2.8 76 *** -0.139 (35.66) (42.39) (36.38) (7 1.11) (64.32) (5 7.06) (3.7 6) (6.11) (5.98) (30.20) (20.7 4) (-1.43) ad j. R 2 0.542 0.555 0.54 7 0.530 0.355 0.196 0.159 0.20 2 0.194 0.45 7 0.210 0.0 21 T st ati st ic s in p ar e n th es es * p < 0.05, ** p < 0 .01, *** p < 0.00 1

(28)

24 Ta b le 3 . Signi fi cant of Di ff e re n ce in D if fe re n ce va lu e a n d t re n d e ff e ct s o n f ou r d im e n si on s Price Eff ec ti ve S p re ad Trad in g V o lu me R etu rn Tre at* Po st (v alu e) ***↓ X ***↓ ***↑ ***↑ ***↑ X **↓ **↓ ***↓ ***↑ ***↑ Tre at* p o st* tr en d ( tr en d ) ** x X *** *** * X *** X *** x *** * p <0.05, * * p <0.01, * ** p <0.001 x: n o t r eje ct t s tati sti cs ↑ po si ti ve c o eff ic ie n t ↓ ne gati ve c o eff ic ie n t

(29)

25

( continued in Page 22)

delisting on daily price of delisting stocks are significantly different from zero.

Hypothesis 1.2 only immediate-term effects of delisting on trend of delisting stocks

daily price is significantly different from zero.

In three different sub-periods , coefficient of Treat are all negative and significant, reported -5.304,-5,440,-4.207 in Table 4 column (A),(B),(C) respectively, which shows the pre-delisting prices of delisting firms are significant lower than pre-delisting price of listing firms. And it is consistent with theory of Bakke, Jens and Whited (2012). As for control group, listing firms show no significantly change in both value of daily price and trend of the daily price through delisting. Specifically, Coefficients of treat*post and tread*post*trend are the results of Difference in difference effect. Coefficients of Treat*post are significantly different from zero in immediate term and long terms, and coefficients of Treat*trend*post are only significantly different from zero in immediate term. So cannot reject the Hypothesis 1.1 and Hypothesis 1.2. For value effect, the price of delisting firms are suffer from significantly decreasing 81% in immediate, reported in table 2 column (A), (G), term and decrease 83% in long term , reported in table 2 column (C), (I). but no significantly decreasing in short term. This no significant estimate in short term seems opposite to existing theory in literatures. As a matter of fact, it is reasonable. In 2011, China mainland stock exchange suffered a bear market. Sun, Tang and Tong (2002) reported that there are co-movements between IPO market and domestic market. So the China stocks listing oversea were influenced by the bearish domestic market in sample year. Besides, the burst of accounting fraud and scandal of China stocks on US exchanges in 2011 and 2012 has a bad influence on other listing China stocks on US exchanges. All the China stocks had a bad performance on US NASDAQ in 2011and 2012. As a result, control group suffer significant price decreasing in short term around delisting, reported in table 2 column (C) decreasing 13.42% and reported in Table 4 column (B) -0.742 , the coefficient of post is significant different from zero. So the significant loss of price

(30)

26

in control group offset some loss of price in treatment group, and the difference between control group and treatment group is not significant in short term. As for trend effect, the trend of price only change significantly in immediate term. It is consistent with Figure 1, daily price finish the dramatically decreasing in about 5 days after delisting. Besides, coefficients of Spread are all more than 16. Compared to coefficient of other independent variables, it is large enough to explain a large part of change in price. So Spread makes biggest contribution to the large loss of price, consistent with the theory of Macey et al. (2008) and Sanger and Peterson (1990).

5.2 Market liquidity

Panchapagesan and Werner (2004) find quoted spreads increase more than double from 12.1 to 33.9 percent, and effective spreads triple from 3.3 to 9.9 percent when stocks are delisted from NASDAQ. I use effective spread to measure the market liquidity. In my study, Hypothesis 2.1 immediate-term, short-term and long-term delisting effects on delisting stocks market liquidity are all significantly different from zero. Hypothesis 2.2 immediate-term, short-term and long-term delisting effects on trend of delisting stocks market liquidity are all significantly different from zero.

In Table 4, coefficients of Treat show that there are significantly difference in pre-delisting effective spread between delisting stocks and listing stocks in immediate- term and short-terms but no significantly difference in long-term. In last fiscal year, there were no significant operating problems, accounting frauds, market liquidity deteriorates and abnormal trading volumes in delisting stocks, otherwise they had been delisted before sample year. So in long-term [-450/+450], there is no significant difference in effective spread between delisting firms and listing firms. As for control group, there are no significant changes in listing firms trough delisting. Both the value and trend of market liquidity have significantly changes through delisting. Specifically, There are 394% increasing in effective spread in immediate term, reported in table 2 column (A), (G); 104% increasing in short term ,reported

(31)

27

The reported in table 2 column (D), (I); 87% increasing in long term, reported in table 2 column (F), (K). Hypothesis 2.1 and Hypothesis 2.2 cannot be rejected. Also, the results are not exactly the same, but consistent with Panchapagesan and Werner (2004), who reported effective spreads triple from 3.3% to 9.9% when stocks are delisted from NASDAQ.

5.3 Daily trading volume

Panchapagesan and Werner (2004) find share volume declines by two-thirds in delisting. In my study, Hypothesis 3.1 short-term and long-term delisting effect on delisting stocks trading volume are significantly different from zero. Hypothesis 3.2 only short-term delisting effect on trend of delisting stocks trading volume is significant different from zero.

In Table 5, Coefficient of ‘Treat’ reports that there are significantly difference in pre-delisting trading volume between delisting stocks and listing stocks in immediate term and short term. While no significant difference in long term. The same implications with market liquidity can explain the no significant difference in long-term. Because in last fiscal year, there were no significant operating problems, accounting frauds, market liquidity deteriorates and abnormal trading volumes in delisting stocks, otherwise they had been delisted before sample year. So in long term, there shows on abnormal and significant difference in pre-delisting trading volume between delisting stocks and listing stocks. As for control group, delisting makes on change on listing firms trading volumes. Harris et al (2008) report that investors react 60 days prior to delisting. Presented in figure 3, the pre-delisting trading volume become abnormally fluctuates in 30 days before delisting. And the post-delisting trading volume decrease dramatically during 10days after delisting. So there is no significant change in trading volume in immediate term .Reported in table 2 column (A),(G), trading volume increase by 30% in immediate term; reported in table 2 column (C),(I), decease by 73% in short term; reported in table 2 column

(32)

28

(E),(K), decrease by 61% in long term. As for trend of trading volume, there is significant change in trend of trading volume in short-term, present in graph 3. In all, cannot reject the Hypothesis 3.1 and Hypothesis 3.2.

5.4 Daily return

In my study, Hypothesis 4.1 immediate-term, short-term and long-term delisting effects on delisting stocks daily return are all significantly different from zero.

Hypothesis 4.2 immediate-term, short-term and long-term delisting effects on trend

of delisting stocks daily return are all significantly different from zero.

Coefficients of Treat*Post are significantly different from the zero in immediate term and long term. Not significant in short term, so Hypothesis 4.1 can be rejected. Treat*post*trend are all significantly different from the zero, which shows the daily trend of daily return change significantly in immediately-terms, short-term and long-term. Hypothesis 4.2 cannot be rejected

Further, the coefficient of Treat*Post is negative in immediate-term, while the coefficients are positive in long-term. It means the return significantly decrease in immediate term after delisting, and then significantly increase in long-term. Higher return in long term of delisting stocks can be explained by the higher risk in OTC exchange. As for trend, delisting stocks have higher volatility in OTC than in NYSE. So those higher risks get the higher risk premium.

6. Robustness checks

In part, aiming to check consistence of two main DID effect estimates. Lu and White (2014) explained adding or removing repressors as a way to examines how certain core regression coefficient estimates behave. According to Bae and Wang (2012). I

(33)

29

used Market capitalization, Debt to assets and Cash as standard control in main regressions. Doing robustness checks with adding additional control variables, I choose to add those existing in previous literatures. Control (1), volatility of return. Park, Lee and W.Park (2014). Control (2), Earning, cash flow and book-to-market. Beavera, McNicholsa and Priceb (2007).

The results of additional regression are in the Appendix. The significant of the core coefficients do not change, and the value of coefficients only have slightly changes.

7.Conclusion

7.1 Conclusions and Implications

My results support the main theory existing in previous literatures, delisting significantly destroy market quality and liquidity. Further, I provide more specific analysis and different conclusions on the real effect of delisting in difference sub-periods. Firstly, daily price of delisting stocks decrease significantly in immediate-term and long-term. Average daily price of delisting firm decrease 81% in immediate-term and 83% in long-term though delisting. The significant loss in price is consistent with previous literatures. However, significant price decreasing in other listing China, making no significantly DID change in Short-term. It is the consequence of the bear market in China and China stock accounting scandal in US. Secondly, market liquidity of delisting firms is significantly destroyed in immediate-term, short-term and long-term. Effective spreads increase near four times in immediately-term, increase 104% in short-term, and increase 87% in long-term. The conclusion are not exactly the same but in line with past theories , Panchapagesan and Werner (2004) reported effective spreads triple from 3.3 to 9.9 percent when stocks are delisted from NASDAQ. Thirdly, trading volume declines by near 30% immediately, but not significant. Because one month before delisting, the trading volume had abnormally increasing and larger fluctuating, which represents react of

(34)

30

investors to delisting risk, Which, in line with Angel et al. (2004) report that investors react 60 days prior to delisting. Besides, Trading volume significantly declines 73% in short-term and 61% in long-term. Panchapagesan and Werner (2004) reported share volume declines by two-thirds. After finishing decreasing in volume in first 10 day of post-delisting. The highly volatility trading volume came down to the normal tend. Fourthly, daily return decrease significantly in immediately-term; Also not significantly decreasing in short term; And increase significantly in s long-term. It is in line with the implication that OTC has higher risk and higher risk premium.

7.2 Limitations

7.2.1 Number of treat sample and Time period choose

Aim to research the real effects on four dimensions during China stocks accounting scandal period, which has the largest number of delisting China firms in such short period since 1990. However, excluding voluntary delisting and data inaccessible firms, only 34 firms left. This number is not large enough, just larger than 30. If increase the sample number I think the new result would be more accurately, although much similar to present one.

7.2.2 Incomplete data of treatment group

In Robustness checks, adding more control variables have not influence my research results. It does not mean control variables are not important. I have got the complete daily trading date. However the financial data of delisted firms in OTC market is incomplete and hard to get. As Control variables in my model are almost financial data come from financial statements, I had some missing data of control variables for some stocks. So I instead the missing value the average year value. If there were complete data, the coefficients estimated would be more accurately.

(35)

31

8.Reference

Bae, K. H., & Wang, W. (2012). What's in a “China” name? A test of investor attention hypothesis. Financial management, 41(2), 429-455.

Bakke, Tor-Erik, Candace E. Jens, and Toni M. Whited. "The real effects of delisting: Evidence from a regression discontinuity design." Finance Research Letters 9, no. 4 (2012): 183-193.

Beaver, W., McNichols, M., & Price, R. (2007). Delisting returns and their effect on accounting-based market anomalies. Journal of Accounting and Economics, 43(2), 341-368.

Clyde, P., Schultz, P., & Zaman, M. (1997). Trading costs and exchange delisting: The case of firms that voluntarily move from the American Stock Exchange to the NASDAQ. The Journal of Finance, 52(5), 2103-2112.

Jiang, Guohua, and Hansheng Wang. "Should earnings thresholds be used as delisting criteria in stock market?" Journal of Accounting and Public Policy 27, no. 5 (2008): 409-419.

Harris, J. H., Panchapagesan, V., & Werner, I. M. (2008). Off but not gone: a study of NASDAQ delistings. Fisher College of Business Working Paper, (2008-03), 005.

Kadlec, G. B., & McConnell, J. J. (1994). The effect of market segmentation and illiquidity on asset prices: Evidence from exchange listings. The Journal of Finance, 49(2), 611-636.

Lu, X., & White, H. (2014). Robustness checks and robustness tests in applied economics. Journal of Econometrics, 178, 194-206.

Panchapagesan, V., & Werner, I. M. (2004, March). From pink slips to pink sheets: Market quality around delisting from NASDAQ. In EFA 2004 Maastricht Meetings Paper (No. 4572).

Park, J., Lee, P., & Park, Y. W. (2014). Information effect of involuntary delisting and informed trading. Pacific-Basin Finance Journal, 30, 251-269.

Sanger, G. C., & Peterson, J. D. (1990). An empirical analysis of common stock delistings. Journal of Financial and Quantitative Analysis, 25(02), 261-272.

Shumway, T., & Warther, V. A. (1999). The delisting bias in CRSP's NASDAQ data and its implications for the size effect. The Journal of Finance, 54(6), 2361-2379.

(36)

32

Appendix

:

Tab le 5. R egr ession s fo r ro b u st n ess ch eck s P an el A: Th e D ep en d en t V aria b le is D ai ly P rice in t h ree d if fer en t term s Imme d ia te – te rm[ -5 /+6 0 ] Sh o rt -t erm [ -6 0/ +6 0] Lon g-term [ -45 0 /+4 50 ] (A) (B ) (C) (D ) (E) (F) (G) (H) (I) Tr ea t -5.5 88 *** -5.4 06 *** -5.3 04 *** -5.7 5 6 *** -5.5 34 *** -5.4 40 *** -4.4 33 *** -4.4 6 2 *** -4.2 07 *** (-10 .96 ) (-10 .68 ) (-10 .4 7) (-11 .50 ) (-11 .1 7) (-10 .9 7) (-4.2 9) (-4.3 6) (-4.0 9) P o st -0.8 05 -0.8 32 -0.8 11 -0.7 1 8 ** -0.7 44 ** -0.7 4 2 ** -0.5 9 6 -0.5 72 -0.5 42 (-0.9 1) (-0.9 5) (-0.9 3) (-3.0 1) (-3.1 4) (-3.1 4) (-1.2 8) (-1.2 4) (-1.1 7) Tr ea t*P o st -4.0 63 *** -3.5 29 *** -3. .1 7 8* ** 1 .36 4 1 .11 8 1 .35 8 -3 .42 9 *** -3 .5 7 0 ** * -3 .31 4 ** * (-4 .82 ) (-4 .59 ) (-4.2 1 ) (0.5 0) (0.1 6) (0.4 9) (-4 .28 ) (-4 .38 ) (-3 .78 ) Tr ea t*T ren d -0.0 30 1 * -0.0 23 2 -0.0 19 7 -0.0 30 6 * -0.0 2 27 -0.0 19 5 0.0 05 4 1 0.0 07 6 7 0.0 14 9 (-2.1 2) (-1.6 5) (-1.4 0) (-2.2 0) (-1.6 5) (-1.4 1) (0.2 2) (0.3 2) (0.6 1) P o st *T re n d -0.2 2 5 -0.2 2 6 -0.2 32 -0.0 34 1 *** -0.0 33 6 *** -0.0 3 37 *** -0.0 44 2 *** -0.0 4 50 *** -0.0 4 57 *** (-0.8 5) (-0.8 6) (-0.8 9) (-5.5 8) (-5.5 5) (-5.5 7) (-4.1 1) (-4.2 2) (-4.2 8) Tr ea t*P o st *T en d 1.2 65 ** * 1.0 7 0* ** 0 .22 0 ** * 0.0 66 4 0.0 75 2 0.0 6 45 0.0 39 8 0.0 38 2 0.0 30 8 (3.9 3) (3.8 4) (3..7 7) (1.0 9) (1.1 8) (1.1 4) (1.0 9) (1.0 6) (0.8 5) St an d ar d Con tr o ls 16 .1 7 *** 16 .95 *** 16 .95 *** 15 .36 *** 16 .32 *** 16 .32 *** 15 .56 *** 16 .5 1 *** 16 .5 1 *** Con tr o l ( 1) (12 1.5 4) (11 8.9 7) (11 9.0 5) (16 7.2 3) (16 2.6 3) (16 2.7 0) (13 5.7 8) (13 2.5 7) (13 2.6 0) Con tr o l ( 2) x x x _c o n s x x x x x x ad j. R 2 0.5 34 0.5 4 1 0.5 42 0.5 46 0.5 55 0.5 55 0.5 38 0.5 47 0.5 47

(37)

33 P an el B : T h e D ep en d en t V aria b le is E ff ect iv e Sp re ad in t h re e d if fe ren t te rms Imme d ia te – te rm[ -5 /+6 0 ] Sh o rt -t erm [ -6 0/ +6 0] Lon g-term [ -45 0 /+4 50 ] (A) (B ) (C) (D ) (E) (F) (G) (H) (I) Tr ea t 0.0 44 6 *** 0.0 40 8 *** 0.0 39 6 *** 0.0 45 0 *** 0.0 40 7 *** 0.0 39 3 *** 0.0 15 5 * 0.0 10 1 0.0 10 5 (15 .44 ) (14 .00 ) (13 .7 8) (12 .8 7) (11 .60 ) (11 .31 ) (2.3 4) (1.5 2) (1.6 0) P o st 0.0 11 0 * 0.0 10 6 * 0.0 10 7 * 0.0 0 40 1 * 0.0 03 8 1 * 0.0 03 7 8 * 0.0 0 2 82 0.0 0 29 4 0.0 0 2 6 2 (2.2 1) (2.1 1) (2.1 5) (2.4 1) (2.2 8) (2.2 8) (0.9 4) (0.9 9) (0.8 9) Tr ea t*P o st 0.2 65 *** 0.2 66 *** 0.2 65 *** 0.1 2 6 *** 0.1 23 *** 0.1 25 *** 0.0 84 7 *** 0.0 81 9 *** 0.0 83 3 *** (19 .29 ) (19 .33 ) (19 .4 7) (24 .61 ) (23 .94 ) (24 .7 0 ) (8.7 1) (8.4 1) (8.6 4) Tr ea t*T ren d 0.0 00 25 0 ** 0.0 00 2 68 *** 0.0 00 2 21 ** 0.0 00 2 60 ** 0.0 00 27 8 ** 0.0 00 2 27 * -0.0 00 34 9 -0.0 00 35 5 -0.0 00 36 3 (3.1 0) (3.3 1) (2.7 7) (2.6 7) (2.8 5) (2.3 5) (-1 .21 ) (-1 .25 ) (-1 .32 ) P o st *T re n d -0.0 00 14 1 -0.0 00 08 88 -0.0 00 16 0 0.0 00 275 *** 0 .00 025 7 *** 0.0 00 2 49 *** 0.0 00 3 58 *** 0.0 00 3 29 *** 0.0 00 3 29 *** (-0.0 9) (-0.0 6) (-0.1 1) (6.4 5) (6.0 1) (5.9 0) (5.2 1) (4.7 8) (4.8 3) Tr ea t*P o st *T en d -0.0 17 1 *** -0.0 1 82 *** -0.0 1 72 *** -0.0 01 87 *** -0.0 01 89 *** -0.0 01 81 *** 0.0 00 4 6 7 * 0.0 00 5 00 * 0.0 00 5 09 * (-4.3 4) (-4 .59 ) (-4.4 0) (-12 .85 ) (-12 .96 ) (-12 .56 ) (2.0 1) (2.1 4) (2.2 1) St an d ar d Con tr o ls x x x Con tr o l ( 1) x x x Con tr o l ( 2) x x x _c o n s 0.0 46 1 *** 0.0 54 4 *** 0.0 5 21 *** 0.0 46 0 *** 0.0 5 50 *** 0.0 5 28 *** 0.0 44 9 *** 0.0 5 38 *** 0.0 5 18 *** (7 1.8 9) (7 4.3 7) (7 1.1 1) (60 .63 ) (66 .83 ) (64 .32 ) (5 3.7 7) (59 .01 ) (5 7.0 6) ad j. R 2 0.5 21 0.5 18 0.5 30 0.3 44 0.3 41 0.3 55 0.1 81 0.1 80 0.1 96

The real effects of Chinese stocks delisting : evidence from a difference in difference model