Journal
of Financial
Economics
36 (1990) 277-188.
North-Holland
Plant-closing decisions and the market value of the firm* David W. Blackwell The I/niLersity of Georgia, Athens, GA 30602, IJ’SA
M. Wayne
Marr and Michael
F. Spivey
Clemson Unir,ersity, Clemson, SC 29632, USA Received
February
1989, final version
August
1990
We investigate the underlying causes and the announcement effects of plant closings. The closings in our sample do not appear related to takeover activity. Instead. they appear motivated by declining firm profitability. Firms announcing closings have lower earnings than market or industry medians: earnings typically improve slightly after the announcement. We find a negative stock-market reaction to plant-closing announcements.
1. Introduction This paper explores the stock-market reaction to a company’s announcement that it is closing a plant. If the discounted value of expected cash flows from the plant is less than the present value of the expected cash benefits of closing the plant, the decision should increase the firm’s value. The market’s
*We are grateful for helpful comments on earlier drafts received from Raj Agganval, William Baranek. Sanjai Bhagat. Bernard Black, Paul Bolster. James Brickley. Charles Brown, John Byrd, Douglas Foster, Stuart Gilson, Mel Jameson. Steven Jones. Robert Kleiman. Robert Lamy. Ronald Lease, Kenneth Lehn. James Lindlev. Robert McCormick. John Martin. William Megginson, Mark Mitchell, Jim Musumeci, Jeffry Netter, Megan Partch (the referee), George Phillipatos, T. Boone Pickens. George Pinches. Annette Paulsen. Robert Rogowski, Richard Ruback (the editor), Ralph Sanders, Barry Schachter. John Schatzberg, Joseph Sinkey, Clifford Smith, Charles Smithson, John Strong, Lawrence Summers, Paul Torregrosa. Michael Vetsuypens. James Wansley, and workshop participants at Clemson University and the University of Georgia. We presented an earlier version of this paper at the 1988 Financial LManagement Association Meetings. We are grateful to James McIntyre, William Bailey. and especially Kurt Barkley for valuable research assistance. Mary Dehner. Patsy Heil, and Nancy Jackson provided valuable editorial assistance. The authors are responsible for remaining errors.
0304-405X/9O/SO3.50
Q 1990-Elsevier
Science
Publishers
B.V. (North-Holland)
reaction depends. however. on both the net present value of the abandonment decision and the information content of the plant-closing announcement. If the market were fully informed about the value of closed plants, one would expect positive abnormal returns at the announcement. presuming the decision to close increases the value of the firm. If, however, investors are not aware that the plant is unprofitable, or do not understand how the closing will affect the firm’s value, share prices may decline after a plant-closing announcement even if the decision has positive net present value. The announcement may also reveal that the firm has fewer growth opportunities. or less favorable prospects for future cash flows. than previously believed; if su, the market’s reaction will be negative. Further, if the market believes that managers are making a poor investment decision by closing a profitable plant, there will be a negative reaction. We can shed light on the broader question of how the market reacts to corporate investment decisions by investigating the underlying causes and announcement effects of plant closings. The evidence from several recent studies of investment decisions has been mixed. McConnell and Muscarella (1985) find that unanticipated increases in planned capital expenditures have a positive impact on the value of the firm and that unanticipated decreases have a negative impact. Statman and Sepe (1989) find a positive market reaction to project terminations. Their sample comprises cases in which managers continue unprofitable operations too long, trying to recover sunk costs. In such situations, they argue. the market reacts favorably when management finally terminates the project, thereby cutting the firm’s losses. They support this interpretation by finding a positive relation between the amount of negative information available about a project before the termination and the abnormal returns associated with the termination. As part of a study of internal restructuring, Brickley and Van Drunen (1990) examine 30 situations in which firms stop operating a unit. They find a significant negative reaction to such ‘unit liquidations’. Share-price declines are greatest when managers state that the liquidation is motivated by a desire to reduce costs or increase efficiency. Since the market would presumably welcome such value-increasing motivations. the authors infer that investors are responding to some other aspect of the announcement: the news of a unit liquidation may convey negative information about the firm’s investment opportunities. We examine several pieces of evidence from a sample of plant closings to gain further insight into the corporate investment decision. First, we focus on management’s reason for closing the plant as discussed in the Wall Srreer Journaf article announcing the closing. Second, we investigate whether plant-closing decisions are related to the firm’s earnings before and after the closing announcement. Third, we investigate whether plant closings are
motivated by takeo\,er attempts. Finally. we measure the market‘s reaction to plant-closing announcements to determine its consistency with ths available information about the closing. The next section of the paper describes the sample. and section 3 analyzes the evidence. The final section summarizes and concludes. 2. The sample We gathered a sample of 3% announcements of plant closings appearing in the Lthll Streer Jo~umI (WSJ) from IYSO and 19&l.’ Sample selection involved two sttps. First. we read the citations in the Hill/ Sfr.~,ef Jo~r-r~l fndr.~ (WSJI) under the heading .plant shutdowns’ for the kears l%O through 1981. Second. \ve used the RiLzfiom/ ll’ew~paper Ida (NNI). an on-line service. to search the WSJI for citations and abstracts of articles containing key words potentially indicating plant closings. When the same plant closing was described in several articles. we used the earliest article to establish the announcement date. Onl) announcements of permanent plant closings are included in the sample. Thus we exclude automobile manufacturers. which often close plants annually to retool for the new model year. For most announcements of auto plant closings. we could not determine whether the plant would reopen. U’e also limit the sample to firms whose common stock is listed on the New York or American Stock Exchange, and included in the Center for Research in Security Prices (CRSP) daily stock returns tile. The sample consists of 773 manufacturin, ~7 firms and 13 mining firms. Closings during the study period were concentrated in hea\>, industries subject to intense import competition. Primary metals firms account for the largest portion of the sample (37 closings). Machinery (except electrical) experienced 31 closings: rubber and plastics. 29; and chemical and allied products. 27. Industries experiencing the fewest closings were tobacco products (2). lumber and wood products (3). and leather (1). 3. Results 3.1. .Il0nngemerrt’s started reason for the plrrrlt closirrg For each announcement in the sample. we read the WSJ article to determine the reason managers gave for the plant closing. u’e classified closings into four groups: (1) Operation trot profitable. its Inryco Inc. construction
For example, Inland products subsidiary
‘A listing of tirmh in the sample is urailahlr
Steel Co. announced that would close its structural
upon request from the authors
280
D. W. Blackwell
et al.. Plant closings. firm
Table Frequency
distribution
I
of 286 plant-closing announcements the closing. 1980-1984.
Reason
c.alue. and takeor.ers
by management‘s
stated
reason
for
1980
1981
1982
1983
1984
Total
Operation not profitable Consolidation of facilities Labor-management dispute Environmental regulations Not reported
47 8 3 1 i
48 3 4 I
71 9
I
3
32 6 3 0 3
12 8 0 0 5
220 33 II 3 IX
Total
67
57
83
44
35
286
fabricating satisfactory
plant in Melrose Park, IL. because profit level’. (WSJ, S/23/81)
I 0
it was not operating
‘at a
(2) Consolidation
For example, Scott Paper Co. announced of facilities. plans to close its Sandusky, OH, operations during the next 18 months, consolidating the business at its facilities in Dover, DL, and Fort Edward, NY. (WSJ, 6/19/80)
dispute. For example, the meatpacking (3) Labor-management United Brands Co. reported that it had not reached agreement concessions with the United Food and Commercial Workers tional Union and therefore planned to close three plants warehouse, idling 1,680 workers. (WSJ, 6/4/82)
unit of on wage Internaand one
For example, Anaconda announced it would regulations. (4) Environmental close a smelter and a refinery because it could not economically bring the smelter into compliance with government environmental and health standards. (WSJ, 9/29/80) In general, managers’ reasons for closing plants convey bad news about investment opportunities or expected cash flows. To some extent, all of the managerial explanations for plant closings suggest that the plant in question may not be profitable. For example, the costs of following environmental regulations or of settling a labor dispute may make it unprofitable to continue operating the plant. Announcements in the ‘operation not profitable’ group, however, typically cite declining demand for the firm’s product, import competition, increasing input prices, or obsolete technology as the reason for the plant’s lack of profitability. Thus the announcements in this group most likely contain new information about the firm’s investment opportunities or expected cash flows. In table 1 we report the frequency of each reason in each year. Managers attributed a large majority (77%) of the plant closings to unprofitable operations. In the early part of the study period, this group grew steadily
Earnings performance
before and after plant-closing
announcements
In this section we examine the earnings of firms in our sample in the years before and after plant-closing announcements. Earnings information bears on an assessment of plant closings in at least two respects. First, poor profitability in prior years may have motivated the firm’s decision to close a plant. Second, a firm’s subsequent earnings performance may provide insight into the efficacy of the closing. We measure earnings as the annual return on common equity (ROE), computed as net income after taxes less preferred dividends divided by total common equity. We obtain the data from the Compustat Annual Industrial file. We measure abnormal earnings in relation to the’market by subtracting from each sample firm’s ROE the median ROE of all firms on Compustat. We measure abnormal earnings in relation to the industry by subtracting the median RUE for firms with the same three-digit SIC code.’ We use the Wilcoxon signed rank test to examine the significance of differences between the performance of the sample firms and the market or industry. In table 2 we report median abnormal earnings for the firms in our sample in the years before and after closing announcements. We find that the ROES of firms in our sample are significantly lower than the market and industry medians in both periods. Results do not vary much across the subsets defined by management’s stated reason for the closing, but are strongest for the subsample ‘operation not profitable’. We also conduct Wilcoxon tests for the difference in the abnormal earnings between the year of the announcement (year 0) and two years after the announcement (year 2) to determine whether changes in earnings after the plant-closing announcement are statistically significant. We find a significant difference (at the 5% level) between earnings in year 0 and year 2 for the entire sample and the ‘operation not profitable’ subsample, using industry earnings as a benchmark. We find no significant differences for the other subsamples when using the industry benchmark. We find no significant differences for any of the subsamples using the market earnings benchmark. Our results suggest that plant closings are associated with declining earnings and that earnings may improve slightly after the closing. Our previous finding
‘Brickley and Van Drunen (1990) use a similar technique to examine earnings changes around internal corporate restructtirings; Meulbroek et al. (1990) use this technique to examine changes in research and development expenditures around corporate adoptions of shark repellents.
Market-adjusted and industry-adjusted median rates of return on common equity (ROE) from two years before through t&o years after a plant-closing announcement classilied by management’s stated reason for the closing.” Market-adjusted Stated reason for closingh Entire
Median Year
sample
-2 -;
0 I Operation not protitable
-1 -;
0 1 Consolidation facilities
of
-1 -;
0
2 Labor-management dispute
-2 -I
0 I 2 Reason not reported
-7 -7 0
ROE%
- LY3 - 3.80 - 7.09 -4.51 - 4.94 - 2.71 -3.4-I - -t.-lO -1.57 -3.x7
Wilcoxon
p-value
Industry-adjusted Sample size
Median
ROE?
Wilcoxon p-value
Sample size
I s,s
-3.x
lxx 18X IX8
-
0.000
188 IYY IX8 188 IYY
I-17 l-17 I17 I-17 I17
-3.32 -3.12 - 3.36 -3.32 - 0.6 I
0.000 I 0.000 1 0.0001 tl.000 I
I-17 I-17 I-17 117 117
-3.11 - 1.72 - 3.13 - 1.33 -5.X
19 IV 19 IY
IY
-3.12 - 6.33 - 0.70 - 3.02 - 3.09
0.0179 O.OOl5 0.0010 0.0025 0.00 I Y
I’) IY 19 IY I9
-6.01 - j.YY - I 1.30 -5.91 - 10.37
Y Y Y 4 Y
- 1.86 _ 5.42 - IO.13 - ;.t7 - 6.69
0.0055 0.0050 O.UO55 0.0107 O.OIJ3
Y Y Y 9 Y
- 0.40 - 4.53 - 3.78 - 7.-17 0.9 I
0.0 I66 l3.0051 0.0066 0.00 15 0.01 16
13
-1.12 - 7.53 - I I.53 - 5.Y2 - 5.99
IXS
I3 I3 I3 I3 I3
3.33 3.Y9 _) 3.32 IX
0.000 I 0.000 1 0.000 1 n.000 I
I 0.000I
13
13 I3 13
“ROE is net income minus preferred dividends divided by common equity. All measurements are made at the end of the previous year. To obtain market-adjusted ROE we subtract the median ROE for all firms on Compustat from the firm’s ROE. To obtain industr)-adjusted ROE we subtract the median ROE for all firms on Compustat with the same three-digit SIC code from the tirm’s ROE. Sample sizes vary because of the availability of Compustat data. hReasons for closing are obtained from the CVdl Stert Joumtrl article announcing the closing.
that managers most frequently cite poor profitability closing is consistent with our findings on earnings. 3.3. Plant closings and takeorer
as the reason
for the
activity
In this section we investigate whether the market for corporate control motivates managers to close plants by determining whether the firms in question were the targets of takeover attempts. Conceivably, managers may
Table Plant-closing
announcements
Pad
A: Vumber
19x0 1981 lY82 I983 198-t
8 16 II 5 5
Total
-18 Panel B: A’umber
Number of years before/after
$ < years before
I
0<
_< f,
I
0 < years after
2 $
4 < years after
5
Total
years after
of )enrs
Number of announcements by targets of hostile takeover” ____ 1 8 Y 3
I
5 2
in our
sample
that
were
Number follouing
takeover
of announcements a takeover attempth 0 0 1 0
I
2
32.
6
before or after rhr plum closing annoutzcemem rakroc er citrempr begrm Number of closing announcements
> 5 years before 3 < years before 5 5 I < years before I 3 years before
3
by the 18 tirms targets.
of closingannow~~mcnt~by target firms in OUT sample
Closing announcements
Year
I <
(1980-lY8J)
that the
Number of announcements by targets of hostile takeobcrs
6 20 II
0 9 6
2
I
3 .;
1
1 1
1 _)
4s
I
22
“This number includes announcements occurring either before or after takeover attempts. “This number includes announcrmcnts by tarpets of hostile and friendly takeovers. ‘None of the tirms in our sample announced a plant clo\ing on the same day that a takro\sr attempt began.
close a plant because of an actual takeover attempt or as part of a major restructuring motivated by a potential takeover attempt. Table 3 contains the results of comparing our firms announcing plant closings with a list of takeover targets compiled by the Directorate of Economic Policy Analysis (DEPA) of the Securities and Exchange Commission. Only 48 (16.8%) of the 256 plant closings in our sample were announced by takeover targets. For comparison we use the results of Mitchell and Lehn (19901, who examine 1,158 industrial firms followed by I/nflle Line to determine whether they were involved in control transactions during the 1980-1988 period. Mitchell and Lehn find that 10.4% of their sample firms were targets of either friendly or hostile takeovers. Only six (12.5%) of the plant closings announced by takeover targets in our sample occurred after the takeover attempt; 37 closings (77%) were announced more than a year before the takeover attempt began. Fifteen of the 22 announcements made
by targets of hostile takeovers came more than a year before the takeover attempt. In short. we find little evidence of a direct connection between takeover attempts and the plant closings in our sample. Only a small proportion of closings involve firms that were takeover targets. Within this subsample. moreover, most closings were announced long before the takeover attempt. Our evidence suggests, therefore, that plant closings are more closely related to declining profitability than to takeover threats.
3.4. 3.41.
Wealth ejjects of plant-closing annow~cemer~ts Statistical procedures
We examine returns over a two-day announcement period: the day before the WSJ report of a plant closing (day 0) and the day on which the article appears (day 1). We use a two-day period because the WSJ report of a plant closing may not be published until the day after the information is released to the public. We use coefficients from the market model to calculate abnormal returns.j Our estimation period is from event day - 170 to event day -21. Data are sufficient to allow us to estimate the market model for 244 plant closings. Once we estimate abnormal returns for individual firms, we calculate cumulative abnormal returns by summing abnormal returns over days of interest. We also calculate average cumulative abnormal returns for various subsamples. We use t-statistics to determine the statistical significance of abnormal returns and cumulative abnormal returns. The standard errors of abnormal returns are based on the estimation-period regression (event day - 170 to event day -21). 3.4.2.
Results for the entire sample and for subsamples defined by reason
In table 4 we report the results of the event study. Over the announcement period (day 0 to day l), we find a statistically significant cumulative abnormal return of -0.55% (t = -3.00). Over day - 1 to 1, the cumulative abnormal return is -0.72%(t = -3.21). Table 4 also shows cumulative abnormal returns for each group of plant closings (except ‘environmental’, which includes only three firms). Abnormal returns are negative for all subsamples, but statistically significant only for ‘operations not profitable’. For this subsample, the cumulative abnormal return over the announcement period (day 0 to day 1) is - 0.59% (t = - 2.72). abnormal return is -0.72% Over day - 1 to day 1, the cumulative ?See Brown and Warner
(1985)
for details.
Table
4
Cumulative abnormal returns (CAR) around the announcement day (denoted announcements of plant closings for the entire sample and for subsamples defined ment’s stated reason for the closing. lYYO- 19X-l. Reason
for closing”
CAR (days 0. I lh
I) for 2-l-1 by manage-
CAR (days - 1. I)
Entire sample Mean r-statistic’ Median Interquartile range Percent negative Sample size
- 0.55?’ - 3.00 - 0.25c; 3.875 54.10 21-l
- o.7?%7c’ -3.21 - O.‘h”c 3.75C; 55.71) 24-l
Operation not profitable Mean f-statistic Median Interquartile range Percent negative Sample size
- O.jYc; - 2.72 - O.‘Y? 3.;1cc 53.00 IYh
- 0.72% - 2.75 - 0.78% 4.33% 51.60 1%
-o.jjc; - Il.00 - 0.2sc; 2.03? 64.05 25
- I.079 - I .34 - 0.91% 2.OY”c ‘6.03 /S
Labor-management dispute Meart I-statistic Median Interquartile range Percent negative Sample size
- O.‘-Icr - 0.27 -0.01~~ 2.12? 55.50 Y
0.02”c 0.01 0.28ci I.1335 U.-t0 Y
Reason not reported Mean r-statistic Median Interquartile range Percent negative Sample size
- 0.135 - 0.55 - O.OYcc 6.13r; 53.so I4
Consolidation of facilities Mean f-statistic Median Interquartile range Percent negative Sample size
- 0.7occ -0.74 0.13% 6.964 53.80 14
“Only three firms in our sample closed plants because of rn~ironmrntal regulations and only one had suficient data for us to estimate a market model. The announcement-period cumulative abnormal returns for other intervals in this table were not statistically signiticant for this tirm. In the interest of brevity. we do not report the actual cumulative abnormal returns and test statistics for this firm. The number of announcements in the other categories shown in this table differs from the numbers given in table I because of the availability of daily stock returns. hDay 0 is the day before the W’uU Strert Jurtmal date. “Significant at the 5% level. ‘The null hypothesis is that the mean cumulative abnormal return equals zero. The r-statistics are based on the standard errors from the estimated regression of the market model (event day - 170 to event day - 21).
(t = -1.75). We believe that our failure to detect statistically significant abnormal returns in the other subsamples reflects the small number of observations. Overall, our event-study results are consistent with two possible arguments. First. plant-closing announcements reveal new information about expected cash flows or investment opportunities. Second. the market believes that managers are making poor investment decisions. We cannot distinguish between these two explanations with our sample and statistical techniques. 3.4.3. Anticipated
l‘ersus unnnticipated
closings
If the market’s negative reaction to plant-closing announcements reflects a response to new information revealed by the closing. rather than the closing itself, then we might expect the reaction to be most pronounced when the
Table 5 Cumulative announced
abnormal returns (CAR) around the CVuN Street Jotrn~ctl date (denoted 1) for plant closings in the period 1980-1981. by the number of closing announcements in
th; two years preceding Number
the year-of
the announcement.
of previous CAR (days 0. 1)
CAR (days - 1. I)
No previous announcement Mean t-statistic’ Median Interquartile range Percent negative Sample size
- 0.85r; - 7.63 - O.OYci 3.595 53.10 80
- 1.21%b - 3.11 - 0.86c7c
One previous announcement Mean t-statistic’ IMedian Interquartile range Percent negative Sample size
- 0.38cr - 0.33 O.SIC; 2.-t?? 3X.00 -12
- 0.2Y% - 0.5 1 0.099 1.0% 16.00 43.
O.SOc; 0.78 0.925 2.77cc 56.20 18
O.Y6% 1.05 1.18% 3.50% 39.00 15
closing announcements”
Two or more previous Mean t-statistic’ Median interquartile range Percent negative Sample size “104 tirms made therefore,
3.87% 63.00 80
announcements
closing announcements before the beginning of our sample only the 140 closing announcements occurring after 1981 are used.
period
(1980):
bSignificant at the 5% level. ‘The null hypothesis is that the mean cumulative abnormal return equals zero. The t-statistics are based on the standard errors from the estimated regression of the market model (event day - 170 to event day -21).
surprise is greatest. If there have been previous indications of poor prospects. a closing announcement may be less likely to evoke a negative response. Accordingly, we tried to determine whether the firms in our sample had a recent history of closing other plants. For each firm announcing a plant closing in 1982 or later (140 closings), we used the CV~zllSrrert Journal fnde.r to determine the number of plant-closing announcements in the two years preceding the announcement in question (e.g., 1980-1981 for a closing announced in 1982). More than half (57%) of the firms that closed plants in 19s: or later had not announced any other plant closings in the previous two years; 30% had announced one earlier closing; and 13% had announced more than one. In table 5 we report abnormal returns for subsamples defined by the number of previous plant closings. For the 130 closing announcements made in 19S:! or later, we find statistically significant cumulative abnormal returns only for those closings that were the first to be announced by the firm in two years. Results were not significant for firms that had recently closed other plants. Small subsample size may be partly responsible for the lack of significance. The first plant-closing announcement. however. perhaps conveys information not fully anticipated by the market, prompting a negative response. Later closing announcements by the same firm. in contrast. may not convey new information to the market.
4.
Summary
We find that the primary reason firms close plants is disappointing profitability. Few of the plant-closing announcements in our sample were made by firms that were targets of takeover attempts. Among firms that were takeover targets. most announced their plant-closings more than a year before the takeover attempt. We find that the earnings performance of firms in our sample is lower than market or industry earnings in the years before the plant-closing announcement. We observe. however. a slight improvement in earnings in the years following the announcement. These results suggest that poor earnings may be more important than the direct pressure of a takeover attempt in motivating plant closings. We find that the stock market reacts negatively to plant-closing announcements. We argue that the negative revaluation may not be tied directly to the closing itself, but. rather to its potential underlying causes: declines in demand or profitability that lead value-increasing managers to reduce the firm’s capacity. Our results are also consistent with the argument that managers are making poor abandonment decisions. In either case, the market receives negative information about firms’ prospects from plantclosing announcements.
References Brickley. James A. and Leonard D. Van Drunen. 1990. Internal corporate restructuring: An empirical analysis. Journal of Accounting and Economics 12. 251-250. Brown. Stephen J. and Jerold B. Warner. 1985. Using daily stock returns: The case of event studies. Journal of Financial Economics 14, 3-31. Meulbroek. Lisa. Mark Mitchell. J. Harold Mulherin. Jetfry Netter. and Annette Paulsen. 1990. Shark repellents and managerial myopia: An empirical test. Journal of Political Economy. forthcoming. McConnell. John and Chris Muscarella. 1985. Corporate capital expenditure decisions and the market value of the firm. Journal of Financial Economics 14. 399411. Mitchell. Mark L. and Kenneth Lehn. 1990. Do bad bidders become good targets‘?. Journal of Political Economy 98, 371-398. Statman. Meir and James F. Sepe. 1989. Project termination announcements and the market value of the firm. Financial Management 18. 7-t-81.