The effect of industrial strikes on the value of shares listed on the Johannesburg Stock Exchange¶
Bhana, N. (1997). The effect of industrial strikes on the value of shares listed on the Johannesburg Stock Exchange. Investment Analysts Journal, No. 44 (Summer). Investment Analysts Society of Southern Africa.
| Type | journal article |
| Scope | Economics and economic history of apartheid |
| Method | econometric |
| Period studied | 1984–1993 |
| Regions | South Africa, Johannesburg Stock Exchange |
| Themes | unions strikes, business and state, methods data sources, labour markets wages |
| Citation key | bhana1997 |
Argument¶
Using an event-study design, the paper tests how the Johannesburg Stock Exchange (JSE) prices industrial strikes. Strike announcements are followed by significant negative abnormal returns, and the losses are not recovered at settlement, so strike costs appear permanent rather than transitory. The market anticipates some of the cost, with a significant fall on the day before the announcement, but the paper argues it underestimates it, since about 70% of the total 30-day decline comes after the announcement. This is read as inconsistent with strong-form market efficiency and as implying a short-selling strategy. The paper also finds that longer strikes bring larger negative reactions, and it draws advice for managers conducting wage negotiations.
Evidence and method¶
Business Day, monitored 1984-93, for strike announcements. Strike start and end dates were verified with the companies and with industrial relations consultants such as the Labour Research Service. JSE share price tapes and the JSE Overall Index were used for the market model. The sample is 50 randomly chosen JSE-listed companies involved in strikes, plus a comparison sample of settlements that averted strikes. Secondary literature on strike costs is mainly US.
Place in the debates¶
- Efficient markets: does the share market anticipate strikes and price their costs correctly? (see Worker consciousness, strikes and conflict on the mines and shop floor). Finds partial anticipation (negative drift before the event, and a reaction that varies with strike length) but systematic underestimation of costs. Concludes the results are inconsistent with the efficient market hypothesis.
- Costs and benefits of strikes (to firms, unions and the economy) (see Worker consciousness, strikes and conflict on the mines and shop floor). Treats strikes as costly negative investments for shareholders and as a lasting loss. Also acknowledges the view that strikes are integral to collective bargaining and can raise productivity.
- Wages, productivity and competitiveness in South Africa (see Worker consciousness, strikes and conflict on the mines and shop floor). Presents employer and union views (Boyd vs Patel) without adjudicating. The study measures only share-price effects.
Findings¶
-
A significant negative abnormal return occurs on the day before the strike is reported in Business Day, suggesting news-wire reaction or leakage. (p. 4)
-0,82% and was statistically significant at the 0,001 level.
-
The paper notes that thin trading on the JSE required beta adjustments. (p. 3)
One of the features of the JSE is that a large proportion of securities are thinly traded.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Company details and consultant verification of strike start/end dates (Labour Research Service and other industrial relations consultants) | interviews oral history | 1984-1993 | Verified start and end dates of strikes and confirmed which companies averted strikes. |
New data¶
Event-study dataset built by the author: a random sample of 50 JSE-listed companies involved in strikes announced in Business Day 1984-93, plus a compiled list of companies that averted strikes, with verified strike dates and abnormal returns over days -30 to +30.
Inference¶
Quantification: Statistical analysis.
Event study: market model (Bowman 1983, with Bradfield-Barr lag adjustment for thin trading) gives abnormal returns and cumulative abnormal returns around strike start, settlement and averted-strike dates; t-tests against zero; comparison of strike vs strikes-averted samples and of strike lengths.
The author's caveats¶
The author notes that the sample (drawn from Business Day) is not representative of all firms engaged in collective bargaining and applies only to large companies. He notes that many JSE shares are thinly traded, which biases beta estimates (addressed with the Dimson and Bradfield-Barr lag adjustment), and that data on strikes averted are not easily accessible. He also notes that the findings may be specific to this sample.
Gaps in the evidence¶
The paper gives no list of the 50 firms, strike dates, industries, or sample sizes for the settlement and averted-strike samples, so the results cannot be checked, and the sample is skewed to newsworthy large firms. It lacks a comparison with fundamentals (earnings, output, wage settlements), and does not control for confounding news such as concurrent announcements or political events. Union records, Department of Manpower strike statistics and company reports existed then and could be used to link share price effects to actual strike duration and costs.
Replicability: Data poorly documented.