Author: Henry Hayes
Graphics: Business Review at Berkeley
Defense contracting is a powerful and profitable industry, where certain times are more profitable than others. The market knows this… but do they know soon enough? Below, this article explores how the share prices of four major U.S. defense contractors responded to five major conflicts over the last 25 years, trying to make sense of each result.
Contracting, Expectations, and Market Return
War! Who is it good for? Northrop Grumman, Lockheed Martin, Raytheon, and Boeing. These firms are, of course, defense contractors; companies who agree to manufacture, procure, and or distribute defense-related goods or services requested by our government. The government assigns a monetary value to each request — or contract — and then presents it to various firms.
These firms subsequently engage in a competitive bidding process to secure the contract, which is ultimately awarded to the lowest bidder. The firm awarded the contract then typically works closely with the government to ensure that the contract’s full scope has been fulfilled. Payout structures and schedules vary from contract to contract, but in all cases, the funds dispersed to these firms come from the defense budget.
Given the role these firms play, it would be reasonable to infer that periods of war are often their most profitable. The stock market, of course, recognizes this, with share prices for these firms often rallying on news of armed conflict, especially if US involvement or arms sales are likely.
Below, this article seeks to explore the degree to which the breakout of armed conflict affects share prices for American defense contractors, and the patterns that may emerge. This will be done by conducting an event study to find the cumulative abnormal returns (CAR) for four different firms (Boeing, Lockheed Martin, Northrop Grumman and Raytheon) following five different war breakouts or war-catalyst events: 9/11 (2001), Russia’s annexation of Crimea (2014), the Saudi Invasion in Yemen (2015), Russia’s invasion of Ukraine (2022), and the October 7th Hamas attack (2023).
This process requires regressing a particular stock’s movements against the movement of the market for a certain window prior to the event — here, 120 days — and finding a Beta and Alpha value from this (these will be discussed further in the Methodology section below). Then, a model must be made of the expected returns of the stock for the year following the event — based on how the market moved that year—and then this must be compared with the actual returns of the stock.
Methodology
Beta (β) is a measure of covariance between a stock and the market; it can be thought of as an expected move of a single stock given a certain move in the overall market. For example, a Beta of 0.5 indicates an expected move of 0.5% for a given stock if the whole stock market moves by 1%. To calculate Beta, we compare the daily returns of a stock with those of the market over a given time period. First, we measure how much each day’s stock return deviates from its average return, and do the same for the market. We then multiply these two deviations for each trading day and sum the results. Finally, we divide this sum by the squared sum of the market’s deviations from its own average return. This ratio gives us Beta, which can also be interpreted as the slope of the trendline when plotting the market’s daily returns (independent variable) against the stock’s daily returns (dependent variable).
Alpha (α), on the other hand, measures the degree to which a stock outperforms the market on a risk-adjusted basis over a given period of time. One way to understand this is by comparing it to the familiar equation of a straight line: y = mx + b. In this analogy, Beta (β) is the slope of the line (m), which tells us how much the stock’s return moves in response to changes in the market (x). Alpha, by contrast, is the intercept (b): it represents the stock’s expected return when the market itself shows no movement. In practice, the stock’s expected return (y) is calculated as (market return × Beta) + Alpha, with Alpha serving as the “extra” return not explained by overall market movements.
Beta and Alpha were calculated for each of the four companies, for the window of 120 days prior to each measured event, yielding a total of 20 values each (five events and four companies). From this, it was possible to plot the following:
- An expected daily log return figure for a given company’s stock in the year following a given event. This is calculated by multiplying the log return of the market by the Beta, and adding the Alpha.
- An actual daily log return figure for the same period.
- A figure known as cumulative abnormal return (CAR), which represents the difference between a stock’s actual performance, and its expected performance given its behavior before the event.
It is important to note, however, that the log return values on the graph are additive over time due to unique properties of log returns. This means that, for example, a log return of 1% each day for three days will appear as a 3% log return on the third day of the measurement period. This is in contrast to how returns actually work — where three days of consecutive 1% gains would amount to a 3.03% gain.
A CAR approach was chosen to measure event response because it controls for the most systematic and measurable source of spurious return variation — overall market returns. (“Spurious” here means variations in market returns not caused by the event we want to measure.) While other firm-specific events can add background noise, market return is usually the largest and most reliable source of such spurious variance, since it reflects broad market conditions that are likely to affect all firms operating in that market.
Findings & Analysis
September 11th Attacks [09-11-2001]
Interpretation
In the year following 9/11, every single company outpaced their expected returns — their yellow lines ending above zero. However, Boeing and Raytheon, who both have extensive exposure to the commercial aircraft industry, experienced sharp downturns for the first few weeks following the tragedy, which makes sense given the nature of the 9/11 attacks. It is interesting to note however that, although these firms were still lower at the end of the year than at the start, their loss was less than that which would be expected considering their previous performance with respect to the market.
Russian Invasion of Crimea [2-20-2014]
Interpretation
Here, we see a completely different outcome: every single firm underperforming their expected returns. This can plausibly be attributed to the fact that the invasion of Crimea, though unexpected, was not a complete surprise. Prior to the invasion, there were contributing factors or signs of a likely conflict. There were increased Russian military exercises along the Ukrainian border, Pro-Russian protests in Crimea, and there was the appearance of “little green men” (soldiers in unmarked uniforms) in and near key sites in Crimea, to name a few indicators. Therefore, it can reasonably be concluded that, in all likelihood, the market responded to these signs prior to the formal annexation, creating periods of outperformance in the pre-event measurement window. This, in turn, made the subsequent expected return figures unusually high, and distorted the degree to which CAR was useful in following the firms’ returns following the event.
Saudi-Arabian Intervention in Yemen [3-26-2015]
Interpretation
Here we find a case very reminiscent of the previous one, with almost all of the firms underperforming their expected returns. Much like the example of Crimea, there was a factor of anticipation prior to this conflict in Yemen. We therefore find it reasonable to conclude that a similar phenomenon occurred here, distorting the expected return values.
Russian Invasion of Ukraine [2-24-2022]
Interpretation
Here, we find what is likely a manifestation of the same phenomenon as the prior two cases. Interestingly however, every single stock finished in the green for that year. Considering this, it may seem puzzling that all of these stocks (except for Boeing) were judged to have underperformed expectations —especially given that 2022 was almost an entire year of bear market conditions, with the market posting -8% returns during the period measured. This mismatch is due largely to the fact that all of these stocks posted very low betas, and quite high alphas prior to the Ukraine invasion, meaning that they responded minimally to the market’s moves, maintaining a steady upward trajectory, and thus the model expected similar behavior going forward (hence the smoother expected return lines, particularly of Northrop Grumman and Lockheed Martin). These low betas heavily suggest conflict anticipation, as stock price movements became largely decoupled from the movement of the broader market, and instead aligned with new developments on the potential conflict.
October 7th Hamas Attacks [10-7-2023]
Interpretation
Here, we see not only some highly unusual graphs, but unprecedented outperformance for every company except Boeing (which was dealing with its own non-defense-related issues at the time). This is likely due to three primary factors: how unexpected and brazen the attack was, high Alpha levels due to nominal outperformance in the pre-event window (due to it coinciding with the war in Ukraine and the associated volatility), and low Betas due (again) to the pre-event window coinciding with the war in Ukraine — another non-market-related, yet extremely consequential return-altering factor.
Conclusion
It seems that conflict is not always a reliable predictor of any outperformance, let alone a certain degree of outperformance for US defense contractors. There were more instances of underperformance of expectations than outperformance. This is not always the case — with 9/11 and October 7th providing two notable examples of stark outperformance for nearly every firm. Though these models are certainly thought-provoking, the degree to which we can draw definitive conclusions from them is uncertain given the extremely dynamic nature of global markets and politics. Some interesting new approaches to such a study could be to potentially extend the timeframe of data captured before the event, hopefully negating the anticipatory effect mentioned. However, this risks including periods of performance including a wider array of unrelated events, potentially clouding our expectations. Another approach could be to look at periods of time before the conflicts themselves broke out to attempt to locate when they began and to what degree market anticipation affected the stock price.
While perhaps unsurprising, it certainly is interesting to see how well the market sets prices given the likelihood of global conflicts; even when other sources may not be as good at this. It would be interesting to explore to what extent this is the case. If the predictive power of markets is sufficient, we might even see more deviations from expected returns during the transition from peace to conflict-anticipation than during the period after the conflict occurs. This would explain the consistent post-event underperformance we see in the models.
Take-Home Points
- War does not guarantee defense stock outperformance. Firms more often underperform expectations than beat them.
- Clear outperformance only followed 9/11 and October 7, 2023; in other cases, results were muted or negative.
- Anticipation matters. Markets often price-in conflict risk before headlines break, raising pre-event baselines.
- Event-study analysis (α, β, CAR) shows post-event returns often disappoint relative to those inflated baselines.
Totally awesome!!!!! Bravo’