The Financial Performance of Public Furniture Firms with Different Positions in Antidumping Actions

Transcription

1 FUNDAMENTAL RESEARCH For. Sci. 64(5): oi: /forsci/fxy017 Copyright 018 Society of American Foresters economics The Financial Performance of Public Furniture Firms with Different Positions in Antiumping Actions Xufang Zhang an Changyou Sun The Unite States has importe a large amount of wooen beroom furniture from China since the early 000s. The antiumping investigation in 003 an aministrative reviews in 010 an 015 have allowe the antiumping uties to be continuously levie on Chinese firms since 005. Domestic firms in the Unite States, however, have been eeply ivie on how to aress the challenge brought by the import growth. In this research, the event-stuy metho is employe to examine the effects of antiumping actions on the financial performance of selecte public furniture firms in the Unite States, with particular attention being pai to their positions on the trae intervention. The analyses reveal that the abnormal returns are mostly negative, except those for the aministrative review in 010. The positions of these selecte public firms in the petition an review inee have impacts on their financial performance. For those firms that are not on the list of government reports, there is a very limite reaction in the equity market. In most cases, the market oes not istinguish between the supporting an issenting firms. Policy implications of these finings for the growth of furniture inustry in the Unite States are iscusse. Keywors: antiumping petition, ARMA, event stuy, GARCH, wooen beroom furniture T here have been increasing imports of wooen beroom furniture by the Unite States in recent years. From 1996 to 016, the Unite States importe a total of $54.6 billion wooen beroom furniture; the average annual import value was $.6 billion, the minimum was $0.6 billion in 1996, an the maximum was $3.7 billion in 015 (Figure 1). Among the 30-plus active exporting countries, China an Vietnam have been the major suppliers. For all the imports over , $16.8 billion or 30% of them were from China, an $13.9 billion or 5% were from Vietnam. Meium suppliers inclue Canaa (10%), Malaysia (9%), an Inonesia (7%). The imports have gaine market shares at the expense of the omestic inustry. The Unite States International Trae Commission (ITC) reporte that the share of import in total consumption was 53% an 78% in 00 an 009, respectively (U.S. ITC 004a, 010). The capacity utilization rate in the omestic inustry has been falling from 73% to 58% over the same perio. The corresponing ata after 009 are not available, but it is perceive that these trens have continue in recent years. In response to the rising imports, several omestic firms an labor unions turne to the feeral government for trae intervention. In October 003, an antiumping petition was initiate against Chinese beroom furniture firms. Since January 005, the antiumping uties between 0.83% to % have been levie on iniviual Chinese firms. In 010 an 015, two five-year aministrative reviews were conucte, an the uty orer has been maintaine. Overall, the investigation an uty levy on Chinese firms have change the trae pattern of wooen beroom furniture. Most Chinese firms face the prohibitive uty rates an were eliminate from the market. In contrast, Vietnam has been continually expaning its export to the Unite States, with the market share of nil in 00 to over 50% in 016 (Figure 1). A number of stuies about the furniture inustry have been conucte in recent years. For example, Drayse (008) analyze the tren of globalization in furniture manufacturing. Several features have been ientifie, incluing global prouction networks establishe by large manufacturers, retailers seeking to reuce costs, an the reuction of trae barriers. Buehlmann an Schuler (009) examine the status of the U.S. househol furniture inustry an conclue that changes in the global economy might create opportunities for omestic manufacturers to regain some losses incurre in the past. More recently, Liebman an Tomlin (015) compare shareholer gains from the Byr Amenment with the losses from subsequent retaliatory measures against the export from the Unite Downloae from by guest on 15 November 018 Manuscript receive September 0, 017; accepte April 13, 018; publishe online July 7, 018. Affiliations: Xufang Zhang (xz35@msstate.eu), Grauate Research Assistant. Changyou Sun (cs58@msstate.eu), Professor, 775 Stone Blv., Department of Forestry, Mississippi State University, Mississippi State, MS Forest Science October

2 Figure 1. Monthly import values (million ollars) of wooen beroom furniture from major supplying countries by the Unite States from January 1996 to April 017. States (e.g., steel proucts an wooen beroom furniture). The analysis reveale that the increase in share returns of U.S. firms receiving subsiies exceee the share eclines experience by firms targete with retaliatory tariffs. More specifically relate to the antiumping investigation on beroom furniture imports in the Unite States, Wan et al. (010a) examine the impact on market competition with an import eman system, an Wan et al. (010b) use intervention analysis to examine the impact on iniviual price an value series of iniviual suppliers. The antiumping investigation showe trae epression effects on China, but trae iversion occurre to Vietnam an other countries. Sun (011) analyze the price interaction between China an Vietnam in the import market of the Unite States. China was foun to be the price leaer, an its price has been evolving more inepenently. Luo et al. (015) assesse the international trae competition of wooen beroom furniture in the Unite States, the Unite Kingom, Australia, an Japan. The antiumping uty against China impose by the Unite States has force China to expan its exports an maintaine strong trae growth in other markets. Overall, these previous stuies relate to wooen furniture have mainly focuse on the aggregate market competition among suppliers at the country level. The performance of iniviual furniture firms, however, has not been examine so far. The objective of this stuy is to analyze the effect of antiumping actions on the financial performance of publicly trae furniture firms in the Unite States, with particular attention being pai to the positions of iniviual firms in the trae intervention. Since 003, the antiumping investigation an two aministrative reviews have generate affirmative ecisions to protect omestic firms. Major events in these actions are all covere in the stuy. A total of 1 publicly trae firms are selecte an analyze. These firms are groupe into four categories: Support, Dissent, Outsie, or All. The event-stuy metho is employe to measure both the abnormal return an volatility cause by the events of interest. The market moel use in classical event stuy is extene with several components to take into consieration the time series properties of financial ata. Theoretically, these events are hypothesize Management an Policy Implications The finings from this research reveal that iniviual events in the trae intervention can have ifferent impacts on the financial performance of public furniture companies. The positions of iniviual firms in the petition an reviews inee have impacts on their financial performance. The significant estimates are mainly associate with the supporting an issenting groups. The results at the firm level are more iffuse than those at the country level. Nevertheless, they reveal the sophisticate competition in the furniture inustry of the Unite States. How to aress the pressure from globalization an import growth will continue to be a challenge to the omestic furniture inustry. One irection of efforts is to continually seek collective trae interventions by U.S. government. Another irection of efforts is that iniviual omestic furniture firms can strive to improve their competitiveness. The global economy has become increasingly integrate over time. Therefore, ultimately it is critical that U.S. furniture firms are competent participants in the global market. Downloae from by guest on 15 November Forest Science October 018

3 to generate positive impacts on the financial performance of the supporting firms, negative impacts on the issenting firms, an no impact on the outsie group of firms. This research will make at least three major contributions to the literature. First, this stuy focuses on the impact of antiumping events at the firm level. Finings from aggregate country-level analyses (e.g., Wan et al. 010b) can reveal little information about the impact of trae intervention on iniviual firms. Thus, the outcomes from this stuy will be more helpful to policy iscussions relate to iniviual firms. Secon, the effects of antiumping actions will be further categorize an compare across firm groups, epening on whether a firm relates to or supports the trae intervention or not. This perspective is innovative an reflects well the reality that omestic furniture firms have been eeply ivie on how to aress the challenge brought by globalization. The results will reveal whether the position of a firm has any iscernible impact on its financial performance. Finally, time series properties of the ata will be thoroughly consiere in the moeling of both return an risk. In the literature of forest economics, event stuy has usually been use with limite attention being pai to time series properties (Mei an Sun 008, Niquiet 008). However, failing to consier time series properties may generate biase results (Corrao 011). In aition, an event can change the risk of a firm, so an analysis of both event-inuce abnormal return an volatility will allow us to have a more balance view of the effects on furniture firms. Methoology Event stuy, also known as event analysis, is a set of statistical moels that can be use to examine the impact of an event on the financial performance of publicly trae firms (Corrao 011). Various techniques of event stuy have been evelope over time, an in recent years, consierable efforts have been mae to consier time series properties of financial ata (Corhay an Ra 1996). Notably, event-inuce volatility has receive more attention along with a stanar analysis of abnormal return. For a comprehensive presentation of the theoretical founation of event stuy, its strength, limitation, stuy esigns, an moel choices, see the reviews by MacKinlay (1997), Corrao (011), an Sorescu et al. (017). As specific applications, event stuy has been use to examine the impact of national elections (Bialkowski et al. 008), floating exchange rates (Chortareas et al. 01), an terrorist attacks (Mnasri an Nechi 016). In forest economics, the metho has been use to analyze mergers an acquisitions (Mei an Sun 008), forest policy reforms (Niquiet 008), an timberlan investment (Piao et al. 017). In this stuy, the impacts of several events relate to the antiumping investigation an reviews on wooen beroom furniture imports are examine. These events cover the perio of 003 to 017. The classical market moel is augmente with several components to consier the serial correlation an time-varying heterosceasticity inherent in financial time series ata. Abnormal returns cause by events are first measure an analyze. Event-inuce volatility is then examine to assess the impact of these events on iiosyncratic risk of the firms. Abnormal Return Analysis In event stuy, a firm s abnormal return cause by an event is efine as the ifference between the actual return an normal return of the firm on a ay within a preefine event winow. The normal return is the expecte return ha the event not occurre. In classical event stuy, a market moel is usually employe to calculate the normal return. The resiuals in the moel are assume to be serially uncorrelate an have constant variance. However, financial ata are often characterize by serial correlation, which can be aresse by the autoregressive moving average (ARMA) moel. Volatility clustering is also a common feature of financial ata, which can be treate by the generalize autoregressive conitional heterosceasticity (GARCH) moel (Engle 198, Bollerslev 1986). As a result, the market moel in event stuy can be extene in several ways to take those into consieration (Murg et al. 016). In this stuy, the ARMA-GARCH market moel is aopte to analyze the impact of investigation an review events. Specifically, the moel has the following mean an variance equations: P Rit = αi + βirmt + φij Rit, j + θik ζit, k + ζit (1) h j=1 Q k=1 = ω + γ h, 1 + ϕζ, 1 () it i i it i i t where R it is the aily return of firm i; R mt is the aily return of market portfolio m; an t inexes a ay within an estimation winow. In this stuy, the Stanar & Poor s (S&P) 500 Inex is chosen as the proxy of the market portfolio. The isturbance term of ζ it has a zero mean an conitional variance h it, i.e., ζ it Ω t 1 N(0, σ ζ = h it it ), where Ω t 1 is the information set available at time t 1. The parameters to be estimate for firm i are α, β, φ j, θ k, ω, γ, an ϕ. The above moel will be estimate over the estimation winow before the event. The with of the estimation winow is specifie as 300 traing ays, which is the same for all the events. The en ate of the estimation winow will vary by event. It will be etermine to avoi the impact of the event, as explaine in etail in the next section. In estimating the moel, the proper length of time lags P an Q (as inexe by j an k) for the ARMA specification is ientifie using the Akaike Information Criterion. The optimal lags shoul aequately aress the serial correlation in the resiuals an also result in a parsimonious moel. For the GARCH specification, the ARCH-LM test can be use to examine if the GARCH effect is present. Preliminary analysis of the ata utilize in this stuy confirms the existence of ARCH effect. In aition, several stuies have compare the performance of various GARCH moels in event stuy, incluing Exponential an Threshol GARCH moels (e.g., Chen et al. 007). GARCH(1, 1) has been prove to have a stable performance with a relatively straightforwar specification, an thus it is aopte in this stuy. The estimate parameters an the observe market portfolio return can be use to calculate the normal return for the firm over the event winow. Thus, the abnormal return can be compute as: ARi = Ri R i (3) where AR i is the abnormal return for firm i on ay, R i is the normal or expecte return, an inexes a ay in the event winow. The length of the event winow can be enote as D= 1 + 1, where 1 an are the beginning an en of the event winow relative to the event ate, respectively. Forest Science October Downloae from by guest on 15 November 018

4 When the abnormal returns are scale or stanarize with their variance, they generally possess better statistical properties (Boehmer et al. 1991, Bialkowski et al. 008). Thus, the statistics erive in Hilliar an Savickas (000) are followe in this stuy. Specifically, the abnormal return can be ajuste by the time-varying variance as follows: SR = AR h ( ) i i i 0.5 where SR i is the stanarize abnormal return for firm i on ay over the event winow, an hi is the estimate variance for ay from the ARMA-GARCH market moel. Abnormal returns for iniviual firms on specific ays can be aggregate across all the firms uner consieration, over time for a given perio, or both. The aggregation allows an overall inference to be rawn with regar to the event of interest when an event affects several firms in the market simultaneously, which is also applicable to this stuy. Thus, the stanarize abnormal return can be aggregate across companies an over time as follows: N i=1 494 Forest Science October 018 (4) 1 SR N = N SR (5) i ( )( ) CSR AR h i (6) i, 1, = = i 1 = 1 N CSRN, 1, = N CSR (7) i=1 i, 1, where SR N is the average stanarize abnormal return for all the N firms on ay ; CSR i, 1, is the cumulative stanarize abnormal returns for firm i over the perio of [ 1, ]; an CSR N, 1, is the average cumulative stanarize abnormal returns for all the N firms over the perio of [ 1, ]. By fixing 1 at a specific number (e.g., 7) an then increasing graually (e.g., from 6 to +7), the evolution of the cumulative abnormal returns over the event winow also can be calculate by ay an emonstrate graphically (MacKinlay 1997, Sun et al. 013). To assess if SR N an CSR N, 1, are statistically significant (Hilliar an Savickas 000), the following statistics can be use: G = SRN ( N 1) SR SR N 0.5 ( i i N =1( ) ) ( ) CG = CSRN N N 1,, 1, ( 1) N CSR CSR i i N =1 (, 1,, 1, ) 0.5 (8) 0.5 (9) where both the statistics have a Stuent t -istribution with ( N 1) egrees of freeom. Abnormal Volatility Analysis In general, the financial risk of a publicly trae firm is compose of systematic an unsystematic parts. In the ARMA- GARCH market moel, they are measure by the magnitue of the estimate coefficient of the market portfolio (i.e., βi ) an the volatility of the isturbance term. In this stuy, we choose to focus on the unsystematic risk an iiosyncratic volatility of the firms uner consieration. With the methoology propose by (Hilliar an Savickas 00), the event effect on the volatility of stock returns can be analyze. A multiplicative abnormal volatility parameter can be constructe to measure the magnitue of the changes in volatility cause by an event for a group of firms. This parameter is efine as: λ N ( ) 1 N N i j=1 i=1 1 i j N j j=1 1 AR N AR = N 1 N ( N ) h + N h (10) where λ N can be calculate for a group of N firms on ay in the event winow. If the parameter value is equal to one, the event has no impact on the unsystematic volatility. A value of the parameter greater than one implies increasing volatility ue to the event, an vice versa. Similar to the analysis of abnormal return, the significance of the event-inuce volatility can also be examine for a given perio within the event winow. To conuct a formal test, the following statistics can be constructe: AV N = λn 1 (11) ( = N 1 1 ) + CAV N, 1, = λ ( 1) (1) where AV N is the average abnormal volatility for a group of N firms on ay, an CAV N, 1, is the average cumulative abnormal volatility for the perio of [ 1, ]. The cumulative abnormal volatility can be ecompose by ay with an incremental increase in. Furthermore, CAV N, 1, can be ivie by the number of ays in the sub-perio to reveal the aily change of volatility. To test if the average or cumulative abnormal volatility values are statistically significant, the following statistics can be employe: Z CZ =( N 1) λ (13) N 1, = 1 N =( N 1) λ (14) where both of them have a chi-square istribution, an the corresponing egrees of freeom are ( N 1) an ( N 1)( + 1). 1 Events, Firms, an Data In this section, the events an ates use in the analysis are efine first, incluing ates an estimation winows by the event. Then, how to etermine an select the firm groups associate with each event is presente. Finally, ata sources for all the information use in this stuy are escribe. Events an Dates In response to the increase imports, several omestic firms an labor unions turne to the feeral government for trae intervention. Since 003, three major actions have been taken, i.e., the antiumping investigation an two five-year reviews. In general, these activities are highly regulate by aministrative law in the Unite States (U.S. ITC 015), an each action has a series of steps that together can last for more than one year. Therefore, there are numerous events associate with the trae intervention on wooen beroom furniture imports from China. Fortunately, only a few of those events are critical to the whole process, an they are selecte an inclue in this stuy (Table 1). Downloae from by guest on 15 November 018

5 Table 1. Events an ates for the antiumping investigation an reviews on beroom furniture imports from China. Date Event information 003 investigation File the petition by the American Furniture Manufacturers Committee Announce the affirmative an preliminary antiumping etermination by DOC Announce the final antiumping etermination by DOC Announce the final antiumping etermination by ITC Publishe the antiumping uty orer by DOC 010 five-year review Institute the first five-year sunset review by ITC an DOC Determine by ITC that the antiumping uty shoul be continuously levie Publishe notice of the continuation of the antiumping uty orer by DOC 015 five-year review Institute the secon five-year sunset review by ITC an DOC Determine by ITC that the antiumping uty shoul be continuously levie Publishe notice of the continuation of the antiumping uty orer by DOC Notes: DOC is the Unite States Department of Commerce, an ITC is the Unite States International Trae Commission. The information is collecte from the reports by U.S. ITC (004a, b, 010, 017b). Detaile statistical results are reporte later for the five ates with. The antiumping investigation was initiate on October 31, 003. The preliminary ecision by the Department of Commerce (DOC) was affirmative an announce on June 4, 004. The final positive ecision was announce on November 17, 004 by DOC an on December 10, 004 by ITC. Since January 005, various antiumping uties have been levie on iniviual Chinese firms. Furthermore, an antiumping uty orer shoul be reviewe after it has been implemente for five years. The first five-year review was institute in December 009 an vote in November 010, an the notice of the continuation of the uty orer was publishe in December 010. For the secon five-year review, it was institute in November 015, an the final affirmative notice was issue in February 017. In total, 11 major events are selecte (Table 1). All these events will be analyze through the event-stuy metho. In specifying the en ate of the estimation winow, the impact of the events shoul be avoie. As there are many events uring the long process of each antiumping investigation or review, the en ate for each of the three major actions is fixe to avoi the pollution of the subsequent events on the parameter estimates from the market moel. Specifically, the en ate of the estimation winow for all the events in the antiumping investigation is October 1, 003, which is 30 ays before the first event (i.e., the petition on October 31, 003). Similarly, the estimation winow ens on November, 009 for all the events in the first review an October 1, 015 for those in the secon review. An event winow is compose of several ays aroun the event ate. In this stuy, several alternative event winows are consiere, incluing 15, 7, an 3 ays (e.g., 1 = 7 an = + 7 for the 15-ay winow). Some asymmetric event winows are also consiere to analyze the cumulative abnormal returns an volatility, e.g., [ 1, ]=[ 7,0], [ 3,0], or [0,7]. Firm Selection After efining the events an their ates, the next choice is to select firms that are affecte by these events. In this stuy, firms in the U.S. wooen beroom furniture inustry are selecte by searching two lists. One list is the firms in the four antiumping an review reports publishe the government agency (U.S. ITC 004a,b, 010, 017b). In total, these reports inclue 83 unique firms. The other list is the financial reports file by firms on the website of U.S. Securities an Exchange Commission ( The Stanar Inustrial Classification (SIC) is use to filter the firms. Wooen beroom furniture firms are covere uner two coes: SIC 510 (househol furniture) an 511 (woo househol furniture, no upholstere). A total of 79 firms are foun through the filtering. Most firms on the two lists are privately hel, e.g., Ashley Furniture. That is mainly because furniture firms are usually small an iversifie in proucts. As a result, only 16 firms are foun to be publicly trae. Four of those public firms were not in the equity market uring the whole stuy perio an i not appear on any list in the government reports either, so they are exclue from the analysis. In the en, 1 firms are inclue in the stuy, as liste in Table. These iniviual firms have experience ifferent business paths in the past several ecaes. To reveal the ifferences, market values of six large firms are plotte in Figure. Specifically, Bassett has improve since 014, an its current market value is higher than that in 003. Hooker Furniture an La-Z-Boy are close to full recovery after a ecae. In contrast, Stanley lost most of its market value since 005. As one of the larger firms in the inustry, Ethan Allen lost more than one-thir of its market value in the past 15 years or so. Furniture Brans International probably is the worst case. Since the late 1990s, the firm has outsource much of its furniture prouction to China, an its market value was over two billion ollars in 00. Furniture Brans International went bankrupt in 013. To achieve the stuy objective, these 1 selecte firms are groupe base on their positions in the antiumping investigation an review reports, an then they are associate with the events uner each investigation or review action. There are 11 key event ates liste in Table 1 an four government reports only. Thus, each firm can have at most four positions, an each position status can be vali for more than one event ate within an action. In Table, the positions are liste in four columns: 003, 004, 010, an 015. The action of antiumping investigation is ivie into two phases: the preliminary phase of the events in 003 an the final phase of the events in 004 an 005. The positions for the two aministrative reviews are recore in the columns of 010 an 015. The position of a firm is labele as (i) a supporting position for a petitioner, (ii) a issenting position when the position of a firm is opposing or unisclose in the government reports, or (iii) an outsie group when a public furniture firm was not liste on the government reports at all. To evaluate the total effect of an event, all the firms also are aggregate into one scenario. Thus, for each event, there are four firm groups, briefly labele as Support, Dissent, Outsie, an All. For instance, for the preliminary investigation in 003, four firms expresse clear support to the petition, i.e., Bassett, La-Z- Boy, Stanley, an Hooker. The two firms in the Dissent group are Furniture Brans who oppose the action an Ethan Allen Forest Science October Downloae from by guest on 15 November 018

6 Table. The list of selecte companies, their market values, an their positions in the antiumping investigation an reviews. who i not isclose its position. Six other firms (e.g., Chromcraft Revington) were not liste in the government report, so they are place in the group of Outsie. Note that Furniture Brans International an Chromcraft Revington were eliste in 013, so the secon five-year review in 015 covers a total of 11 firms only. Data Sources Several ata sources are utilize in the stuy. Monthly furniture imports by the Unite States from 1996 to 017 are ownloae from the interactive tariff an trae atabase that has been Market value N Firm name Ticker PERMNO Bassett Furniture BSET La-Z-Boy LZB Stanley Furniture STLY Hooker Furniture HOFT Ethan Allen Interiors ETH Furniture Brans Intl. FBN N 7 Chromcraft Revington CRC O - O N 8 Kimball Intl. KBAL O O Flexsteel Inustries FLXS O O O O 10 Leggett & Platt LEG O O O O 11 Natuzzi S P A NTZ O O O O 1 Select Comfort SCSS O O O O Notes: The permanent number (PERMNO) an market value for a firm are from CRSP (017). The position information is from the reports by U.S. ITC (004a,b, 010, 017b). The firms are classifie into three groups an labele as: Support (+), Dissent (-), an Outsie (O). In aition, all the firms are combine to create a new group as All. Two firms were eliste in 013, an they are labele as N. Uner the position, 003 covers the event ate of , 004 covers the five event ates in 004 an 005, 010 covers the three ates in the first review, an 015 covers the three ates in the secon review, as liste in Table 1. Figure. Daily market values (billion ollars) from 001 to 016 for selecte beroom furniture firms. Position maintaine by U.S. ITC (017a). Uner the Harmonize Tariff Scheule (HTS) system, the wooen beroom furniture proucts mainly refer to HTS The import values by the Unite States are collecte for each of the major supplying countries. These ata are use in the escriptive analysis only. The secon source is the atabase of the Center for Research in Security Prices (CRSP 017). For each firm, its aily return, closing price, an staning shares are ownloae. The price an share ata are use to calculate market capitalization for escriptive analyses only. The return of S&P 500 inex is the proxy for the market Downloae from by guest on 15 November Forest Science October 018

7 portfolio, an it is ownloae from this atabase too. All the aily return ata are expresse in percentage, an they are use to conuct the propose event stuy. Stock return ata in the CRSP atabase is upate once a year. Thus, the ata for 017 was not available when the ata for this stuy was ownloae. To have a complete coverage of the perio of the secon aministrative review, the return ata for these public firms in 017 are ownloae from the Yahoo Finance ( To assess the quality of the ata, the aily returns from the CRSP an Yahoo Finance were compare for each iniviual firm in the year of 016. As a result, the Yahoo ata are foun to be the same an satisfactory. Results All the 11 major events associate with the antiumping investigation an review actions are evaluate through the event-stuy metho. The analyses generate a large number of results. To facilitate presentation, five event ates are selecte from Table 1, an the results are reporte an analyze here. These ates represent the impact of the actions well. The results for other ates are either less significant or ominate by those reporte here. Results of Abnormal Returns Let us first focus on the abnormal return results of two event ates. The first example ate is December 10, 004. That was the ate when ITC announce the final an affirmative etermination with regar to the antiumping investigation. As shown in Table 3, there are four significant estimates of the average stanarize abnormal return ( SR N ) over the 15-ay event winow. For example, the value is 0.75% on the fifth ay before the event for the supporting group of firms. Table 3. Average stanarize abnormal returns (SRN ) an average abnormal volatility ( AV N ) uring the antiumping investigation. Event ay () SR N Support Dissent Outsie All Support Dissent Outsie All ** ** 0.38** ** * ** * 0.71* * ** ** 1.0* * ** *** 13.83*** * ** *** *** * * **.33** 3.1*** * ***.3** 4.47***.6*** *** ** 3.70*** * * 1.13* *** ** ** * ** *** 5.01*** ** * ** * * * * * *** ***.61*** * AV N Downloae from by guest on 15 November 018 Note: *, **, an *** enote the significance at the 10%, 5%, an 1% level, respectively. The four groups are efine in Table. Stanar errors are omitte because of space constraints. Forest Science October

8 When the above abnormal returns on iniviual ays are aggregate over a perio within the event winow, they can reveal if the abnormal returns can persist over time. In Table 5, the average cumulative stanarize abnormal returns for the event ate of are significant in several instances. For the wiest event winow consiere, the estimate, i.e., CSR N, 7,7 = 0.55%, is highly significant for the supporting group, an similarly, 0.74% for all the firms as a group. For narrower perios, the estimates are mainly significant for those before the event ay, i.e., 0.84% for [ 7,0]. Consistent with the estimates of average abnormal return, the cumulative abnormal returns are not significant uring any event winow for the outsie group. The abnormal return results for December 10, 004 are also shown in Figure 3 over the 15-ay winow for all the firms as a group. While significant estimates of SR N occur on two ays only for this group (i.e., 4 an 3), the cumulative estimates ( CSR N, 1, ) are significant on every ay after = 4. Thus, for this scenario, the negative abnormal return is persistent when it is accumulate over time. The secon example ate is January 10, 017. That was the ate when ITC vote an mae the key ecision of maintaining the uty orer uring the secon five-year review. As reporte in Table 4, there are nine significant estimates of SR N, with three being positive. In particular, the issenting group has a large an negative estimate, i.e., SR N, 3= 1.90%. As expecte, the market interprete the affirmative ecision of maintaining the uty orer as ba news to the issenting group an respone with a big negative return before the event ate. The market later mae some correction with a positive return on the event ay. Finally, the results for this example ate over the perio of [ 7,7] are also plotte in Figure 3. In sum, for the secon aministrative review, there are some small positive returns for the supporting group, large negative returns for the issenting group, an moerate negative returns for all the firms combine. Interpreting the results for all the event ates together reveals several patterns. The signs of abnormal returns vary across the major actions. The antiumping investigation in 003 an 004 generate mainly negative abnormal returns. The first five-year review in 010 prouce all positive returns, an the secon five-year review in 015 has both positive an negative returns. Thus, the financial market perceives the antiumping investigation an the secon review more negatively, but in contrast, more positively for the first review in 010. The magnitues of abnormal returns also vary by event. The absolute values of several significant estimates are bigger than 1%, mainly for the events of (i.e., the preliminary ecision in the investigation) an (i.e., the secon five-year review). In particular, the largest estimate of average stanarize abnormal return is SR N, 3= 1.90% for the issenting group uring the secon five-year review in 017. The positive abnormal returns have moerate values, with the largest one being 0.79% only. Overall, the sizes of the negative returns are bigger than those of the positive returns. By firm position, the outsie group is least affecte by all the events. In fact, for the outsie group, no average or cumulative abnormal returns are significant for three out of the five events reporte here. For the few significant estimates, they are all negative. The supporting group has the most significant estimates, but Downloae from by guest on 15 November 018 Figure 3. Average stanarize abnormal returns ( SR N ) an average cumulative stanarize abnormal returns ( CSR N, ) 1, for two selecte event-group scenarios over the 15-ay event winow. (The significance levels of the estimates are inicate by the symbol sizes an shapes.) 498 Forest Science October 018

9 Table 4. Average stanarize abnormal returns ( SR N ) an average abnormal volatility ( AV N ) uring the two aministrative reviews. Event ay () SR N AV N Support Dissent Outsie All Support Dissent Outsie All ** ** * ** * *** 0.79* ** * * *** ** 5.58*** ** * ** *** *** ** 1.75* 1.33* *.16** **.54** * * *** 1.65* Note: See Table 3 for notations. contrary to the expectation, these affirmative ecisions generate negative impacts on their returns for most events. The only persistent positive return for the supporting group occurs uring the first review in 010. For the issenting group, the significant estimates are mainly negative, which is consistent with expectation; the exception is that the first five-year review in 010 also generates some positive impact on the abnormal returns. Results of Abnormal Volatility To elaborate the results from the abnormal volatility analysis, two event-group scenarios are also utilize. The first example is the issenting group on (i.e., the preliminary ecision ate). For this scenario, six out of 15 estimates of the average abnormal volatility ( AV N ) are significant (Table 3). All of them are positive, with the value between 1.13 to Thus, the volatility has increase up to almost ten times on some iniviual ays for the issenting group. Some of the estimates are negative, with the potential of reucing the volatility, but none of them are significant over the 15-ay event winow. When the average abnormal volatility is accumulate over a specific perio within the event winow, it reveals if the impact is persistent over time or not. In Table 5, the average cumulative abnormal volatility ( CAV N, 1, ) is highly significant for all the nine perios consiere in this stuy. For example, the estimate is 0.59 for the 15-ay perio, inicating an average volatility increase by 137% per ay (i.e., 0.59 /15=1.37). In Figure 4, the volatility measures are plotte over the 15-ay event winow for the issenting group on The average abnormal volatility on the thir ay (i.e., 9.80) before the event ate is most visible, followe by the secon largest value of 3.1 on the event ate of The average cumulative abnormal volatility increases graually from about 10 to 0 when the event winow is wiene. The volatility change per ay uring the event winow ecrease slightly over time, but most of the significant values are larger than 100%. The secon example is the issenting group on (i.e., the secon aministrative review). After the event ay, the estimate of average abnormal volatility is significant every ay for this event-group scenario (Table 3). The largest value of occurs on =1. When these volatility estimates are aggregate over a perio within the event winow, the results reveal that the average accumulate abnormal volatility is increasing an significant whenever the ays after the event ate are covere (Table 5). In Figure 4, the large volatility estimate on the first ay after the event announcement shows its ominance an persistence over the whole 15-ay perio. Combining all the event results together, the signs of the significant estimates of abnormal volatility are all positive. Some estimates are negative (e.g., uring the first aministrative review), but none of them is statistically significant. Thus, the events have generally increase the risk of these public firms. Furthermore, the magnitues of abnormal volatility vary by event. The largest impact on volatility occurs on (i.e., the preliminary ecision), Downloae from by guest on 15 November 018 Forest Science October

10 Table 5. Average cumulative stanarize abnormal returns ( CSR N, ) 1, an average cumulative abnormal volatility ( CAV ) N, 1, by event. Perio CSR N, 1, CAV N, 1, [ 1, ] Support Dissent Outsie All Support Dissent Outsie All [ 7, 7] * ** [ 3, 3] [ 1, 1] * [ 7, 0] [ 3, 0] [ 1, 0] [0, 1] * 0.71** [0, 3] ** [0, 7] ** [ 7, 7] *** 0.59*** 16.31*** 16.31*** [ 3, 3] *** 15.67*** 5.97*** 9.37*** [ 1, 1] ** 4.46*** [ 7, 0] 0.74** *** 1.38*** 10.4*** [ 3, 0] *** *** [ 1, 0] * 3.77*** * [0, 1] * ** 3.90*** [0, 3] *** 5.57*** 5.9*** 6.66*** [0, 7] *** 8.4*** 3.53* 6.53*** [ 7, 7] 0.55*** ** ** 0.76*** 5.5*** [ 3, 3] ** 3.58* [ 1, 1] ** [ 7, 0] 0.84*** 1.0*** *** *** 3.16 *** [ 3, 0] 0.78* 0.88 ** *** ** [ 1, 0] 0.76** * [0, 1] ** [0, 3] * 4.6** [0, 7] ** 4.73** 3.83**.06 * [ 7, 7] * * [ 3, 3] [ 1, 1] [ 7, 0] [ 3, 0] * [ 1, 0] [0, 1] [0, 3] [0, 7] * [ 7, 7] 1.36** 1.35* *** 18.01*** 0.68*** [ 3, 3] 1.59* * 1.63*** 9.96*** 1.67*** [ 1, 1] 0.89** ** *** [ 7, 0] 1.07* *** 5.80** [ 3, 0] * 1.11*** 5.81** [ 1, 0] [0, 1] ** *** * [0, 3] *** 4.35** 13.83*** [0, 7] 0.65** *** 1.41*** 1.6*** Notes: See Table 3 for notations. followe by (i.e., the final etermination by ITC) an (i.e., the secon aministrative review). The ates for the preliminary an final eterminations uring the investigation in 004 are critical, so it is within the expectation that they increase the volatility ramatically. The volatility impact is minimal on (i.e., the petition ate), an it is entirely insignificant on (i.e., the first aministrative review). The association between a firm s position an the estimate of the abnormal volatility is consistent with prior expectation. The outsie group is least impacte, an the issenting an supporting groups are most affecte. The magnitues of the estimates for the supporting group an issenting group are particularly large in some cases, e.g., AV N,3 = 6.79 for the supporting group on an AV N,1 = for the issenting group on These results reflect that several of these firms ha volatile market performance uring the stuy perio (e.g., the bankruptcy of Furniture Brans International in 013). Summary an Discussion With accelerating globalization, wooen beroom furniture inustry in the Unite States has face growing pressure because of the importation from China in the 000s an Vietnam in the 010s. Antiumping uties have been continuously levie against Chinese firms since 005, but the overall performance of the omestic inustry has still been weak. The omestic inustry has been Downloae from by guest on 15 November Forest Science October 018

11 Figure 4. Average abnormal volatility ( AV N ) an average cumulative abnormal volatility ( CAV N, ) 1, for two selecte event-group scenarios over the 15-ay event winow. (The significance levels of the estimates are inicate by the symbol sizes an shapes.) eeply ivie on how to aress the challenge. In this stuy, the effect of antiumping actions on the financial performance of publicly trae furniture firms is analyze, with particular attention being pai to the positions of these firms in the antiumping petition an reviews. The classical market moel is extene with several components to consier time series properties of financial ata, an both event-inuce abnormal return an volatility are examine. Several outcomes are generate from the event stuy, an they have important implications for the evelopment of furniture inustry in the Unite States. These outcomes can be summarize by the event, return, risk, an a firm s position. Overall, iniviual events can have either positive or negative effects on the financial performance of the selecte firms. The effects on abnormal return an volatility o not necessarily match each other for a specific event. Some events mainly generate abnormal returns while others cause abnormal volatility. The abnormal returns are mostly negative, while the abnormal volatility is positive an increasing for all the interventions. Furthermore, the positions of iniviual firms in the actions inee have affecte their financial performance. For the outsie group, the reaction in the equity market is very limite. This may reflect the fact that those firms in the outsie group are iversifie in various furniture sectors, with wooen beroom furniture being one of their proucts only. In comparison, the significant estimates are mainly associate with the supporting an issenting groups. However, the ifference in the equity market responses is small between the supporting an issenting groups. The leaing implication of these finings is that while the overall market prospect of the furniture inustry is gloomy, the performance of iniviual firms can still iffer ramatically. Previous stuies have mainly focuse on the aggregate effect at the country level of the antiumping actions on import values or prices, trae epression effect, or trae iversion effect (e.g., Wan et al. 010a,b). In contrast, the finings from this stuy clearly reveal the heterogeneous performance of iniviual firms at the micro level in face of the import growth pressure. The positions of iniviual firms in the trae invention are relate to their business concentration an strategies. When the overall market conition is not favorable to the furniture inustry, most firms have been negatively impacte, but some have been performing well over time. These results at the firm level are more iffuse than these from the aggregate country-level analyses. Nevertheless, they reveal the sophisticate competition in the furniture inustry of the Unite States. Future stuies can be conucte to examine in etail the ifferences Downloae from by guest on 15 November 018 Forest Science October

12 among these furniture firms to reveal what factors have contribute most to their success or failure. How to aress the pressure from globalization an import growth will continue to be a challenge to the omestic furniture inustry. One irection of efforts is to continually seek collective trae interventions by U.S. government. The finings from this stuy reveal that in most situations, the financial performance of both supporting an issenting furniture firms has been negatively impacte by the antiumping actions against China. Apparently, the market recognizes that trae iversion has occurre to Vietnam, an the effect of levying antiumping uties against China is largely nullifie as long as the import growth from Vietnam is unaresse. However, the possibility of taking a similar roun of antiumping actions against Vietnam in the near future seems low. This is because trae interventions through antiumping actions are costly in both time an financial resources. The effect of antiumping uties against a few selecte countries is generally limite an temporary because of the accompanying trae iversion to non-targete countries. Furthermore, antiumping actions may also invoke retaliatory policies from traing partners (Blonigen an Bown 003, Liebman an Tomlin 015). At present, the pressure on omestic firms from Vietnamese furniture imports is increasing. Therefore, furniture firms an government agencies may nee to continually search for other tools that are both effective an consistent with omestic an international trae policy an law, especially within the legal framework of the Worl Trae Organization. Another irection of efforts in face of import growth pressure is that iniviual omestic furniture firms can strive to improve their competitiveness. The global economy has become increasingly integrate over time. Therefore, ultimately it is critical that U.S. furniture firms are competent participants in the global market. Several previous stuies have analyze possible strategies for improving the competitiveness of the inustry through surveys or qualitative reviews (e.g., Gazo an Quesaa 005, Wan an Bullar 008, 009). General strategies that have been iscusse inclue enhance mass customization, more prouct innovations, clearer customer targets an pricing, more intensive marketing, an better government supports. In the future, these stuies can be expane to cover firms in other countries an analyze what U.S. firms can learn from their growth an success. Finally, it shoul be note that the conclusions from this stuy nee to be interprete with caution. One main limitation of the stuy is that the firms inclue in the analysis are publicly trae companies. While these firms are large an influential in the inustry, they are only a subset of all omestic firms. Most wooen beroom furniture companies in the Unite States are privately hel, with very limite financial information being isclose to the public. In aition, many mergers an acquisitions have occurre in recent years, an the interactions among these furniture firms are worthy of analysis. The change an evolution of omestic furniture prices in the Unite States also have receive limite attention. These research issues can be aresse when new methos or ata sources are available. Literature Cite Bialkowski, J., K. Gottschalk, an T.P. Wisniewski Stock market volatility aroun national elections. J. Bank. Fin. 3: Blonigen, B.A., an C.P. Bown Antiumping an retaliation threats. J. Int. Econ. 60(): Boehmer, E., J. Musumeci, an A.B. Poulsen Event-stuy methoology uner conitions of event-inuce variance. J. Fin. Econ. 30:53 7. Bollerslev, T Generalize autoregressive conitional heteroskeasticity. J Econometrics 31: Buehlmann, U., an A. Schuler The U.S. househol furniture inustry: Status an opportunities. For. Pro. J. 59(9):0 8. Chen, M.-H., S. Jang, an W.G. Kim The impact of the SARS outbreak on Taiwanese hotel stock performance: an event-stuy approach. Int. J. Hospit. Mange. 6(1):00 1. Chortareas, G., A. Cipollini, an M.A. Eissa. 01. Switching to floating exchange rates, evaluations, an stock returns in MENA countries. Int. Rev. Fin. Analy. 1: Corhay, A., an A.T. Ra Conitional heteroskeasticity ajuste market moel an an event stuy. Quart. Rev. Econ. Fin. 36(4): Corrao, C.J Event stuies: a methoology review. Account. Fin. 51: CRSP The atabase of the Center for Research in Security Prices. University of Chicago Booth School of Business, Chicago, IL. Drayse, M.H Globalization an regional change in the U.S. furniture inustry. Growth an Change. 39():5 8. Engle, R.F Autoregressive conitional heterosceasticity with estimates of the variance of Unite Kingom inflation. Econometrica. 50(4): Gazo, R., an H.J. Quesaa A review of competitive strategies of furniture manufacturers. For. Pro. J. 55(10):4 1. Hilliar, J.E., an R. Savickas On stochastic volatility an more powerful parametric tests of event effects on unsystematic returns. Athens, GA. 50 p. Hilliar, J.E., an R. Savickas. 00. On the statistical significance of event effects on unsystematic volatility. J. Fin. Res. 5(4): Liebman, B.H., an K. Tomlin Worl Trae Organization sanctions, implementation, an retaliation. Empir. Econ. 48(): Luo, X., C. Sun, H. Jiang, Y. Zhang, an Q. Meng International trae after intervention: The case of beroom furniture. For. Policy Econ. 50(1): MacKinlay, A.C Event stuies in economics an finance. J. Econ. Literat. 35(1): Mei, B., an C. Sun Event analysis of the impact of mergers an acquisitions on the financial performance of the U.S. forest proucts inustry. For. Policy Econ. 10(5): Mnasri, A., an S. Nechi Impact of terrorist attacks on stock market volatility in emerging markets. Emerg. Mark. Rev. 8: Murg, M., M. Pachler, an A.C.M. Zeitlberger The impact of analyst recommenations on stock prices in Austria ( ): Evience from a small an thinly trae market. Central Euro. J. Operat. Res. 4(3): Niquiet, K Revitalize? An event stuy of forest policy reform in British Columbia. J. For. Econ. 14(4):7 41. Piao, X., B. Mei, an W. Zhang Long-term event stuy of timber real estate investment trust conversions. For. Policy Econ. 78(1):1 9. Sorescu, A., N.L. Warren, an L. Ertekin Event stuy methoology in the marketing literature: An overview. J. Aca. Mark. Sci. 45(): Sun, C Price ynamics in the import wooen be market of the Unite States. For. Policy Econ. 13(6): Sun, C., M.M. Rahman, an I.A. Munn Ajustment of stock prices an volatility to changes in inustrial timberlan ownership. For. Policy Econ. 6(1): U.S. ITC. 004a. Wooen beroom furniture from China (731-TA-1058 final). Publication U.S. International Trae Commission, Washington DC. 40 p. Downloae from by guest on 15 November Forest Science October 018