Political Exposures: Can Insurers Benefit from Data Related to Campaigns?
By Jim Weiss
There are risks in politics—as well as risky politics—and political campaigns with culminating elections may often carry costs for the public that can’t be seen from the voting booth. Before the arrival of “deep data” and predictive analytics, many insurers were largely blind to potential risks introduced or heightened by political campaigns—risks that can range from business losses to personal property damage.
But in the age of social media, when a well-timed tweet has the potential to “go viral” and jump-start a campaign, new tools are emerging that can help bring some of the potential risks from politics into clearer focus.
At the moment, presidential campaigns and election fervor are gripping the nation. Just log in to social media or turn on the television and the viewer is bound to see the coverage of politicians crisscrossing the nation before the voting public. For property/casualty insurance companies, elections may present different risks than whatever reforms enacted by an administration. As campaign staff, volunteers, media, supporters, protesters, and other members of the general public flock to meet and greet candidates, they may leave behind a wake of challenges and potential risks that insurers are left to address. Insurers that review such election cycles with strategic eyes may also see related opportunities to mitigate losses, price more accurately, and even develop new and more relevant products.
Drilling Down on Risk
When drilling down on the subject, the hazards of political elections can become real and observable. A campaign rally, for example, may cause a traffic jam that in turn produces a series of fender benders. Local businesses may experience greater foot traffic from larger crowds when a candidate visits, leading to accidents such as slip-and-falls. National media arriving to cover campaigns may be unfamiliar with local traffic patterns, leading to congested roads. And volunteers going from door to door can cause property damage. Such effects are not necessarily observable at more than a micro level, but examining them can illustrate a larger analytical approach that can be generally applied to various fleeting risks with similarities in that typically they require timely data and methods to understand and act upon.
Under a purely empirical approach to risk assessment, insurance ratemaking bears similarities to the political campaign wherein every step is measured by the latest polling figures. That is, rates are likely to play “catch up”—they may increase once the costs of election-related damages are observed and incorporated (sometimes without even direct attribution to the election) and then phased out until the next election, when the same cycle resumes. A more strategic and less disruptive approach may preemptively build the costs of elections into rate analysis over time based on the related risk factors.
Consider a predictive model that evaluates geographic risk for personal and small commercial policyholders. For example, ZIP codes or census block groups housing diners, bookstores, speaking halls, or state fairs may present elevated risks during election years. Expand the view from a local diner to a national party convention and the degree to which the campaign process may elevate risk becomes more apparent. Today’s cutting-edge approaches to geographic ratemaking are well equipped to deal with these realities by considering risk factors including the following: local business points (candidates’ offices), traffic density and patterns (average vehicle traffic in an area at a given time of day), and local population characteristics (fields of employment, such as campaign workers).
Incorporating the statistics into predictions—in the form of longer-term averages—may help stabilize an area’s rate fluctuations that may largely be attributed to election year damages. Longer-term averages may also offer greater attribution to risk dynamics driving the cost, for example, one might observe to the incremental effects of campaign offices appearing and disappearing from an area every four years rather than just the loss fluctuations.
Social Metrics: Counting Votes and Tweets
Gaining deeper insights into the hazards related to political campaigns also carries the potential to benefit handling of claims by insurers. If campaign events and related phenomena can be correlated with insurance losses, why not prepare adjusters accordingly? Research from Indiana University suggests that analyzing social media metrics such as “tweet share” helped correctly predict 403 out of 406 Congressional elections, so there seems to be no shortage of timely information about politics being communicated via social media. With this in mind, insurers might consider setting up alerts for a high number of politically themed tweets that come from a particular geocode. These tweets may signal a political event under way and suggest that a claims professional should be at the ready. Such an approach falls within the growing discipline of social media event detection. Aside from providing timely service, this would help mitigate the costs of claims and possible postelection rate changes.
Some insurers are also considering novel ways to blend the risks associated with politics into predictions. Presently, Election Day is not a federal holiday, and individual voter information cannot legally be used to set insurance rates in the United States. But in a perfect world in which everyone has an equal opportunity to vote, insurance discounts could be used to encourage voting along with other socially beneficial behaviors. At least one insurance company has reportedly already registered a patent to use voter registration and voting history to estimate risk, purportedly under the assumption that regular voters may be less likely to file an insurance claim. Voting-related discounts may also prove to be an opportunity for policyholders in offsetting higher rates they may receive from living in areas that are frequented by campaigns and may thereby be subject to the related exposures.
Better Ways to Develop Election Coverages
Despite paying continued homage to its grassroots origins, the modern political campaign for even minor political positions has evolved into an intricate event that can involve relatively large numbers of employees and volunteers. Some insurers have capitalized on this opportunity by crafting equally intricate policies to address some of the perceived campaign risks. There have long been ISO commercial multiple peril classifications to specify rates to cover, for example, political campaign headquarters, although more tailored offerings have begun to emerge in the marketplace that blend elements of coverage from general liability, property, auto, employment practices, liquor liability, and crime.
Additional coverages may soon find their way into political campaign-related insurance coverage offerings—and with greater frequency. Cyber coverage will likely become a larger component because campaigns may possess potentially sensitive donor data. Political entities also can make representations online, any one of which could cause a personal advertising injury claim that may result in liability or adjustment expenses. Campaign leadership may also seek coverage for errors and omissions relating to their perceived roles in the outcome of campaigns, for example, if a high level staffer negligently misses a deadline for his or her candidate to appear on the primary ballot in a particular jurisdiction. By examining the range of a campaign’s potential coverage needs, insurers may be envisioning a prime example of the commercial package of the future.
The variety of coverage options connected to political campaigns can only continue to grow for personal lines, as well. According to a Pew Research Center report, online political contributions grew by approximately 50 percent between the 2008 and 2012 presidential elections. Independently, a crowd funding site has developed a coverage option Perk Insurance, which reportedly refunds “micro-investors” for projects that do not materialize, subject to certain specified conditions. To the extent the average contributor to a campaign may be considered a “micro-investor,” the insurer that develops an affordable option to indemnify contributors to failed campaigns may have a solution that could also increase participation in the democratic process. The potential market for such a product would conceivably be up to one-quarter of the U.S. population (that is, those who make contributions).
Further, is it possible that individuals are exposed to “political risk” beyond the value of their contribution? Corporations have been able to procure political risk-related insurance to protect themselves from financial impacts of political developments (e.g., revolutions) in foreign jurisdictions where they conduct business. If an individual’s Social Security or health benefits are affected by elections and the candidate(s) supported are not elected, is there a similarly insurable interest? These questions arguably stretch the boundaries of speculation but provide food for thought in new personal lines coverages.
When personal and commercial policyholders “vote” on their insurance companies at the outset of each renewal period, the issues tend to be much simpler than those discussed on the political campaign trail. Policyholders may be “voting on” some of the following questions:
- Is the price commensurate with my risk?
- Is the policy’s price stable from year to year?
- How soon will the insurer be there after a claim?
- Is there a coverage offering available to meet unique and evolving needs?
The intersection between insurance and the political process remains a fine and fascinating line—one that can provide valuable insight into how insurers can use data and ingenuity to win over policyholders.
Jim Weiss, FCAS, MAAA, CPCU, is director of analytic solutions at ISO, a Verisk (Nasdaq:VRSK) business.