Why do corporations buy insurance?

Yesterday I wondered:

Why do corporations by insurance for fire damage and such?  It seems to me that maybe the oughtn’t, since the cost of insurance is greater than the expected payouts (due to administrative costs, asymmetric information, moral hazards etc).  Investors should presumably prefer corporations to be pure bets, and reduce risk and volatility by holding suitably diversified portfolios.

Today my colleague Peter Taylor, who worked in the insurance industry for many years, replied (reproduced here with permission):

Corporations certainly do buy insurance against fire and very good value it proves to be for them I must say when a large-scale fire does occur.  Your argument was adopted by some large corporations going "self-insured" or creating  their own "captives" but generally it takes one large loss and they are back in the insurance market.  Moreover, the argument for self-insurance can be about saving a few pennies off expenses rather than assessing the real risk – a recent example was Hull Council deciding to self-insure with its own fund against flood rather than pay the market price – underestimating the losses by an order of magnitude.  The reversion to the insurance market is partly to do with shareholders’ wish for stable results as well as their reluctance to accept bad luck.  Shareholders don’t seem to accept that accidents/fires/whatever happen and blame the management (Napoleon’s unlucky generals) so from a management point of view it is much easier to buy the insurance year on year and avoid getting caned when a loss does occur.

I’m still not sure I completely understand why insurance is bought. It might be that shareholders are biased (which seems to be what Peter suggests).  If so, is this a recognized failing? Do sophisticated institutional investors also prefer that the companies they own stock in buy fire insurance?

If so, is there an alternative explanation other than that they are biased?

(a) Maybe shareholders don’t want management to be able to blame poor results on bad luck due to fires etc?  But it seems relatively easy to check if accidents of this sort were responsible for big losses – easier, e.g., than to check whether competition has been unusually difficult.

(b) Maybe shareholders prefer to make certain pure bets:  whether company X has a good business plan, for example, rather than whether company X will also be lucky with regard to fires, floods, etc.

(c) Maybe the amounts spent on insurance are so small that shareholders just don’t care.  But in some cases I would think insurance costs are substantial, e.g. for shipping firms or transport of hazardous materials. Do such companies not buy insurance?

(d) Maybe costumers and suppliers want to know that the company will not suddenly go bankrupt because of a plant fire.  But I doubt that these constituencies ever check how much fire insurance a corporation has bought.

(e) Maybe when making long-term plans, it is valuable to reduce uncertainty as much as possible.  So a corporation might insure against big fires for the same reason that it hedges against currency risks.  But hedging currency risk may not be as expensive as insuring against plant accidents (given thick, efficient currency markets).  And where capital markets are sufficiently efficient, couldn’t a previously profitable firm with good prospects simply raise more capital if it unexpectedly needed to rebuild a burnt down factory?

Or is loss aversion bias partially responsible?

GD Star Rating
Tagged as:
Trackback URL:
  • michael vassar

    I think that insurance companies do two things. They spread around risk, but they also reduce it by, for instance, insisting that companies take prudent measures to avoid fires. Companies increase efficiency by dividing labor among specialists. This includes outscourcing fire prevention.

  • Constant

    Maybe specialization. If insurance is a service, then it may make as much sense to outsource it as it makes to outsource many other services and products. Most companies do not manufacture their own staples, chairs, or light bulbs, deliver their own products using vans that they manufactured themselves from metal that they mined themselves, provide their own utilities, and so on. Companies also often hire outside companies to provide janitorial services and also physical security against intruders. By the same token, then, it may make sense for companies not to provide their own fire insurance, earthquake insurance, and so on.

  • Constant

    If a company cannot meet its business commitments, many more people are affected than just the investors. Insurance may help to insure that a company can keep its business commitments. That is presumably a good thing because it will presumably encourage people to enter into mutual agreements with the company.

    A company that loses a shipment of goods, if that shipment is not insured, will have its revenue stream interrupted. It may be unable to pay its employees, suppliers, and so on, as agreed. Aside from inconveniencing those employees, suppliers, etc., it may indirectly lead to the failure of the business, as employees leave, suppliers sue and/or lose confidence, etc. Thus a problem of a significant but limited scope may balloon into a much bigger problem.

    It may therefore not be sufficient merely for an investor to diversify his investments. While the diversification will limit the loss to him to some fraction of his total investment, the uninsured loss to him will (in the scenario described above) still amount to his share of the entire company rather than merely his share of the shipment. He would prefer to lose only the value of his share of the shipment rather than the value of his share of the entire company. This is true no matter how diversified he is. And the only way to do this may be for the company to do something that prevents the loss of a shipment from becoming the loss of the entire company. And one way to do this is to insure the shipment.

    The company can still, of course, self-insure. But if this self-insurance merely means “if the shipment is lost then we’re screwed”, then such self-insurance will not do the job, which is to prevent a lost shipment from being the end of the company. The self-insurance will have to be something else, such as (for example) setting aside, in cash, an amount equal to the value of the shipment. That is expensive to do, and makes the prospect of buying insurance relatively more attractive. Rather than each company setting aside a large amount of money to insure itself, it may be more economical for many companies to pool their risks, which allows them to set aside a smaller total amount of money (since the probability of simultaneous disasters at all companies is very small). But once you have entered into this territory, you’re doing something that insurance companies specialize in. There’s that word: specialize. A toy manufacturer may prefer to let somebody else worry about pooling risks, and so may the investors behind it.

  • tylerh

    The rule is simple:


    Even if a given risk doesn’t appear to be potentially fatal, one should never forget:


    The day after your warehouse burns down is when your rival will launch an aggressive campaign to grow market share. Having top management focused on defending a product liability lawsuit means no one is “minding the store,” so the product delivery schedule slips, alienating customers. Better to carry liability insurance and have the insurance company’s veteran lawyers do most of the work.

    In short, it’s rational to accept a small, predictable loss to make sure that a deep pocketed friend is legally obligated to help when things get ugly.

    Additionally, most business insurers engage in “loss prevention” — working with the insurance buyer to minimize the risk directly. So buying many forms of business insurance includes buying into a “best practices” consultancy.

  • ed johnson

    I’ve long wondered about this. And it’s hard to talk about it with anyone, because I get the sense that most people who aren’t economists don’t even understand the argument (including your colleague Mr. Taylor).

    I think Constant is on to something; it is important for economic efficiency that the firm be protected from large risks that could cause bankruptcy or major disruption. This would predict that companies would generally have high deductibles, and would only carry enough insurance to prevent major financial disruption. Is this true?

    Some other half-baked ideas:
    (1) Officers of the company who can’t diversify their risks away like general investors can. It’s important to keep these people happy, so buying insurance is an easy and low cost way to insulate these people from irrelevant risks for which moral hazard is a minor problem.
    (2) Self insuring means either keeping a large horde of cash around, or having access to a line of credit. Having too much cash lying around creates an agency problem that firm officers are likely to spend it on bad projects. And perhaps lenders are reluctant to lend to firms after they’ve experience large losses.

    A related puzzle is why companies hedge things like commodity prices. Southwest airlines has famously hedged against increases in fuel prices, much more than other airlines, and it has paid off handsomely for them in the last few years. But it just as easily could have gone the other way, and hedging isn’t free.

  • Stuart Armstrong

    I think Constant is on to something; it is important for economic efficiency that the firm be protected from large risks that could cause bankruptcy or major disruption.

    Immediate bankruptcy isn’t the issue (investors would solve that issue by diversifying – indeed, if the liabilities are much higher than the assets (such as with a dangerously defective product) then bancrupcy is a boon to investors).

    But a major disruption is (or a drawn-out death spiral). Raising cash after a major disaster is hard. Suppliers and customers are worried, and relations are disrupted. Banks see this worry and the disruption, and lack reliable information that the company is still profitable – hence they charge higher risk premiums for their cash. Stockholders desert the company, and the stock price plunges, complicating other ways of financing the future of the company (this is building on Constant’s point).

    There is also a rational-irrationality going on – people know that stockholders and banks will behave irrationally, threatening the future of the company. Therefore they join the stampede, for this irrationality makes the company rationally less viable. Chicken and egg: which came first, the irrationality or the rationality?

    There is also a seperate issue, to do with the principal agent problem. Ed identified one, but there are others: Executives profit more (through reputation and compensation) from great successes and from avoiding great failures, than they generally do at very small marginal increases in profit. After a disaster, an executive with the foresight to buy the correct insurance has his reputation greatly enhanced. One who cancelled the insurance program soon before a major disaster will have his reputation destroyed. These effect dominate the small cost of the insurance.
    Hedging has similar incentives, with the added bonus that the company can become much more profitable after a good hedge, further enhancing the earnings and reputations of the executives.

    But all is not gloom and doom – hedges and insurances are highly liquid, while company assets are not. So insurance could promote greater efficiency in pricing than the company doing its own risk assessements.

  • http://cumulativemodel.blogspot.com aaron

    The missed calculation is opportunity cost. A business has an area of specialization where it earns a good return, but outside of that, it doesn’t get good results. When a disaster happens, the cost isn’t simply the cost of the damages. It’s the impact on cashflows and the cost of liquidating assets or taking on additional debt to get the cash to restore assets. All while enduring a disruption of income.

    I thought the same thing (still do, kind of). But I then realized that it may be possible that insurance is of value to the rich too, since the cost of insurance may be less than return that the person could earn from the cash value of the assets insured.

  • http://cumulativemodel.blogspot.com aaron

    Is the cost of insurance less than the cost of capital for the expected payout?

  • http://cumulativemodel.blogspot.com aaron

    Constant is pretty much right, but not addressing what was positted. The idea is that the company is sitting on cash greater in value than it’s operating assets.

    He is right to avoid addressing this scenario though, since it generally doesn’t exist. The company will generally payout the cash to shareholders or reinvest it. Companies tend to reinvest. It’s arguable that the company should keep the cash not reinvested to cover its insurance liability rather than pay it out to the shareholders, but then you need to refer to my previous question.

    And also, there is the specialization thing. Insurance companies are able to earn a small, basically risk-free return on the cash it holds, covering some of the cost. The company wouldn’t be able to do this or wouldn’t be able to it as well.

  • Tom B

    Combination of principal-agent problem and risk aversion. Managers are risk averse with respect to the company going bust. Shareholders are risk neutral. But managers get to decide on insurance cover.

  • http://buildgeneralai.blogspot.com/ Lars Pensjö

    I think it is all a matter of risk calculation.

    When you want to invest your money into something, you should first decide how high risks you want to take. Generally, you can expect higher returns on investments with higher risks. But only on average.

    It is the same for a company. The owners request informaton about how high risks the company is dealing with. All risks that are “big enough” has to be addressed.

  • Phil Groce

    All businesses already “self-insure” in their own business domain; they assess, undertake and manage at least some of their own risk, or else they wouldn’t be worthwhile investments. Diverse investment portfolios, fuel and other resource hedges, and redundant suppliers/distributors are all forms of insurance against risk.

    External insurance is for domain-independent, “cross-cutting” risks (e.g., fire or theft). The specialists in assessing, undertaking and managing these risks are insurance companies.

    Another hypothesis is that companies externally insure to _insulate themselves from bias_. A third-party adjustor will assess premiums with cold-hearted detachment, but a desire to use the money elsewhere might lead a company officer to underestimate the holdback required, even unwittingly. To paraphrase: a self-insured company has a fool for an insurance adjustor.

    To test these hypotheses: do fire insurance companies self-insure their offices against fire? If so, they are showing confidence in their ability to manage risk in their domain. If not, perhaps there is something to the use of an external agent to assess those risks. Or perhaps they’re simply not a very good insurance company. 🙂

  • http://dhyb.blogspot.com Andy Wood

    Chapter 6 of David Friedman’s Law’s Order deals with this problem. Scroll down to the section subtitled “Moral Hazard: Bug or Feature?”.

    To summarise his argument:
    1. There is a principal-agent problem to be overcome. The shareholders bear the risk of a fire destroying a factory, but the managers control whether appropriate precautions are taken. A manager could have an incentive to cut costs by skimping on fire precautions. The cost cutting would look good when the manager’s bonus is assessed, but the skimping is concealed from the shareholders.
    2. Therefore, the shareholders need to monitor the managers’ actions.
    3. The shareholders, have no specialist knowledge in fire prevention, but insurance companies, who have an incentive to minimise costs due to moral hazard, do.
    4. Therefore, a corporation that hires an insurance company to monitor the managers, transfering the risk to those doing the monitoring, could well be a better bet than one that self-insures.

  • Alan Gunn

    A lot of companies buy liability insurance not for the insurance protection itself but as a way of hiring someone to handle claims, arrange for defense, etc. I know of one large company whose liability insurance had so large a deductible that it had never been exceeded. This seems similar in principle to the services like fire protection that you get with fire insurance. One would think, though, that if this is the main reason for buying fire insurance, deductibles would be large. Is that the case? Or perhaps it’s necessary in the fire-insurance case to put a fair amount of risk on the insurance company so that it has an incentive to take its fire-protection duties seriously.

  • http://www.blog.greenideas.com botogol

    It’s not that hard.
    Corporations should self-insure for losses that they can afford (eg employee drops laptop into the bath)
    Corporations need external assurance for losses that they can’t afford (corporate HQ burns to the ground)

    Just like individuals, really: you should insure your house, but not your bicycle.

  • ed johnson

    Lars, it is only undiversifiable risk that are supposed to earn higher returns, in other words the very type of risk that it’s hard to insure against.

    So we have two competing classes of explanations:
    (1) Buying insurance raises the expected profit of the firm (by preventing expensive disruptions, or by providing monitoring).
    (2) Buying insurance lowers the expected profit of the firm (and the variance in profit), but officers or other large shareholders of the firm choose to buy it anyway (because they can’t diversify, or because their performance is poorly measured).

  • http://www.blog.greenideas.com botogol

    Need to consider the expected *volatitily* of the returns, not just the level. If the 1 in a 100 bad year happens next year, you won’t be around to see the 98 good years.

  • Pat

    turn Modigliani Miller on its head – it’s either because of taxes, contracting costs or investment decisions. Hedging lowers expected tax liabilities by reducing the volatility of income. Tax receipts are like an option the government owns on a company’s assets. Volatility increases the value of an option, so companies want to make that reduce the cost of that option the government has on them.

  • Bob

    Pat has the right idea – buying insurance is negative expected NPV for *everyone* (when insurers know what they are doing) from a premium-payout perspective, so the answer has to be a violation of MM assumptions. I’d second Constant and focus on expected bankruptcy costs – not the disruption that occurs in financial distress (this is diversifiable from a value perspective) but the cost(s) that an unusually high probability of distress creates for the company today. Customers, suppliers, employees, etc. cannot diversify as easily as investors and will generally value stability (they generally do not benefit from upside volatility). Ask yourself, would you book an absolutely critical trip next month on an airline that is rumored to be suffering large losses and running out of cash? When the possibility of financial distress has undesirable consequences on a business, costly insurance can easily pay off.

  • http://profile.typekey.com/bayesian/ Peter McCluskey

    The principal-agent problem is an important part of the effect.
    I doubt any big investors intend to influence decisions whose expected value are this small. But they may unintentionally cause risk aversion as a byproduct of using earnings predictability as an indicator of company quality. (I use revenue predictability as an indicator of good investments with only vague ideas about why it works).
    For the kind of hedging that Southwest airlines does, I think a customer desire for relatively stable prices (which probably involves both biases and a rational desire for ease of comparing prices over time) leads to customers rewarding Southwest for keeping its costs (and therefore prices) stabler than its competitors.

  • Douglas Knight

    Re: Southwest
    I don’t know what they actually did, but it seems logical to me to hedge demand when doing capital expansions.

  • http://profile.typekey.com/robinhanson/ Robin Hanson

    I’ve been hanging out at a finance department here in Australia the last week and actually asked a lunch group this question a few days ago. Andy is right that insurance companies can better monitor some activities. But more important, Constant is right that the disappearance of a company disrupts business flows and is a real cost. There are many sunk fixed costs that are also lost when a company goes. So this justifies some risk aversion. But agency failure could still make actual insurance levels be far in excess of what is justified by this risk aversion.

  • Jonathan

    There is a very standard explanation for rational insurance purchase by a corporation(not that it explains all insurance purchases by a means). Internaly generated funds are cheaper than external funds to the firm. Thus, the point of insurance is to provide cash flow at exactly those times when the firm would have to turn to external sources. The point of insurance is to align cash flows with cash needs. When my plant burns down and prices rise, that’s exactly when I need cash to rebuild.

    See for example http://www.nber.org/papers/4084

  • ed johnson

    If the “disruption in service” explanation is correct, it implies a defect in credit markets. Otherwise you should just be able to borrow (or sell stock) to cover your losses. After the fire (or whatever), either there are still a bunch of positive NPV projects available, or there aren’t. If there are, someone should be willing to fund them. If there aren’t, you should close up shop anyway, insurance or no.

    So what are these credit market imperfections, and why do they exist?

  • ed johnson

    I see Jonathan already answered my question, along with providing a great reference.

  • Konrad

    Shareholders don’t run corporations — they hire management for that. If you’re looking for biases, that’s where you should focus.

    To quote from scripture:
    “We will get hit from time to time with large losses. Charlie and I, however, are quite willing to accept relatively volatile results in exchange for better long-term earnings than we would otherwise have had. In other words, we prefer a lumpy 15% to a smooth 12%. Since most managers opt for smoothness, we are left with a competitive advantage that we try to maximize.”
    From the book of Buffett, year 1995, Ch 4, Verse 12

    It is discussed in many more places throughout these holy scriptures, but the short story is that shareholders don’t generally analyze things that closely. One big loss draws far more scrutiny than continal small insurance payments over the years. Even though risking the loss may be the intelligent decision, managers have a great dis-incentive to make intelligent choices that will guarantee the occasional large loss. Job security wins over a slight gain to the shareholders.

  • http://cumulativemodel.blogspot.com aaron

    To self insure, a company needs to hold highly liquid assests or maintain a workforce and build over-capacity. To do this the company must borrow more, either by issuing stock or debt (or not investing in additional income generating growth or paying out to investers). Therefore, if the cost of insurance is less than the marginal cost of capital for the company, it makes more sense to buy the insurance.

  • Pat

    This place is biased towards attributing everything to bias.

    Let’s compare 2 companies. Company A makes $200 one year and loses $100 the next year because of some disaster they didn’t insure against. Company B makes $50 both years (insurance costs them 150 bucks but they get paid 150 in the event of a disaster). Tax rate is 40%.

    Company A pays .4*200 year 1 and no taxes year 2. $80 total.

    Company B pays .4*50 + .4*50. $40 total.

    Insurance makes tax liabilities less volatile which helps shareholders – if you paid negative taxes when you have negative income, then you wouldn’t need it.

  • http://irs.gov guy in the veal calf office

    There may be a legal & commerical compulsion. All lenders, many vendors, all landlords, and many government agencies (not just workers comp) require minimum levels of insurance and may want to be named co-insured. They don’t want to examine a service provider’s net worth, liquidity, etc. so they just use a simple requirement of a minimum level of coverage.

    Also, your question assumes that a corporation shields shareholders from the corporation’s liabilities. A plaintiff can “pierce the corporate veil” and sue shareholders by proving up that the corporation wasn’t run as a real corporation ans one important factor in that analysis is whether the corp had sufficient funds or insurance to cover its potential lossed.

  • http://profile.typekey.com/nickbostrom/ Nick Bostrom

    I hadn’t expected that this question would generate so many thoughtful responses!

    The two main *new* ideas that I gleaned from the discussion are (a) to the extent that the insurer provides not just insurance but also expert monitoring, that can clearly account for the phenomenon; and (b) and if management holds large stakes in a company that they cannot diversify out of, they may opt to insure more than shareholders would prefer. Also, if earnings stability has a signal value, as McCluskey suggests, that too could be a factor. In addition to these, there is the possibility of lack of perfect capital markets, which was alluded to in my original list of hypotheses.

    As for bankruptcy costs, it seems that this would be a factor that depends on imperfect capital markets. If it were easy for a good company with excellent long-term prospects to raise more cash to make up for a big one-off loss, it wouldn’t go bankrupt and hence would not need to insure for that reason.

  • Paul Gowder

    Simple moral hazard: they have private information about the riskiness of their own behavior?

  • http://magicdragon.com Jonathan Vos Post

    Everything above is vague, intuitive, and sadly qualitative. Perhaps you might start with Claude Shannon’s publications on Mathematical Portfolio Theory. Or Malliavin caculus with applications to stochastic differential equations, by Marta Sanz-Sole, Fundamental Sciences, EPFL Press, Lausanne, distributed by CRC Press, 2005. Or Paul Malliavin and Anton Thalmaier, Stochastic calculus of variations in Mathematical Finance, Springer, 2006. Or ask my frequent co-author Full Professor Philip V. Fellman, Oh, of course. Eliezer Yudkowsky hypothesizes that Professor Fellman does not exist, despite his offical web page:


    Mr. Yudkowsky also denies the existence of Sir Arthur C. Clarke’s long time friend and webmaster John Sokol, while ironically screaming abuse at him over the telephone.

    Professor Bostrom, out of respect to Oxford, can you please keep your autodidact fellow Mr. Yudkowsky from defaming my coauthors, and myself? If only out of academic protocol and collegiality?

    You might also suggest that he telephone the Caltech registrar to verify that I at least existed in 1973 when I earned my first two degrees there. Philosophy should have some routine fact-checking, in the interests of rationality, should it not? Otherwise it is all just language games.

    Thank you in advance for your attention and consideration.

  • Daniel Schiffer

    Most corporation do receive tax refunds when they incur losses by using tax carryforwards and carrybacks.

  • John Markowitz

    I believe the main reason is that corporations are primarily concerned with getting obstacles out of the way so they can do business. Many probably only use 1 travel agency. Sure they could save money by having some admin go online and search Orbitz, Hotwire, etc – but it is simpler and easier to just get it done so they can get back to the business of producing and selling widgets.
    The same goes for insurance. Yes they could spend the time and energy to evaluate their risks and adopt the appropriate self-insurance, but for most companies the savings would not outweigh the hassle.
    Should my widget company develop its own VOIP system, self fund a health care plan for my workers and keep an in-house group of attorney to handle all lawsuits? The answer depends on the scale of my operations. Ford has its own financing arm, but it would be silly for some $40 million dollar company to do it. There may be biases, but in general I think the realities of a free market do a pretty good job in tearing them down.

  • http://localhero.biz shawfactor

    Your analysis ignores the fact that company’s have obligations to debt holders and employees. Unlike equity these groups aren’t diversified and do not benefit from the potential upside associated with self insuring

    Those lending money to a company would likely charge the company a much higher interest rate to cover the extra risk involved with the company being uninsured. Likewise employees may require higher wages.

    Given these additional costs it might be just more efficient to insure.

  • http://profile.typekey.com/tim_tyler/ Tim Tyler

    Insurees are often required to insure by law – usually when someone could be badly damaged or die, and bankruptcy proceeds might not adequately cover the payout.

  • Pocha huntas

    Living it up; living advertisement statesman n a lot more.http://www.pinterest.com/bubblegumcastin