This is an action by plaintiff Steven J. Glusband, as receiver for Michael Starbuck, Inc. and Associates ("MSIA"), to recover losses suffered by MSIA under a bond insuring against the dishonest or fraudulent conduct of employees. Magistrate Francis tried the case before a jury that found appellant Insurance Company of North America ("INA") liable under that bond for the losses. Because we conclude that the losses were, as a matter of law, not caused by dishonest or fraudulent conduct as defined in the bond and were trading losses explicitly excluded from coverage, we reverse.

Michael Starbuck was the sole owner, director, and officer of defendant Michael Starbuck, Inc. ("MSI"), an investment management corporation organized in 1977 and registered as an investment adviser pursuant to the Investment Advisers Act of 1940, 15 U.S.C. § 80b-1 to -21 (1988). In late 1977, Starbuck created MSIA as a limited partnership, with MSI as the general partner, to invest in securities.

Starbuck induced investors, among whom were friends and relatives, to part with their money on false promises of MSIA's following a conservative investment strategy. Instead, Starbuck's strategy was reckless, often involving trading in naked options. To this highly risky activity was added Starbuck's inexperience in such matters, and MSIA's losses mounted until its insolvency brought trading to an end. These losses were temporarily concealed by Starbuck's misrepresenting to his customers the condition of their accounts. Although Starbuck's trading was extremely risky and his statements to customers were highly misleading, there was no evidence whatsoever that Starbuck either converted, or intended to convert, his customers' funds--including any gains from trading--to his own use. Starbuck's sole gain was thus entirely in the form of salary or commission.

In early 1978, Starbuck's insurance broker secured from INA a brokers' blanket bond insuring MSI in the amount of $500,000 beyond a $500,000 deductible. The terms of this bond are the central issue on this appeal.

In January 1980, Glusband was appointed as receiver for MSI and MSIA and filed the present action on their behalf against INA and other defendants in December 1980.1 The case was tried to a jury before Magistrate Francis. The jury returned a verdict against INA in the amount of $500,000, and INA appealed.

The bond affords coverage for loss through any "dishonest or fraudulent act" by employees of MSI. Two provisions are pertinent to the breadth of this coverage. The first defines the term "dishonest or fraudulent act." The second excludes from coverage "trading losses." Whether viewed separately or together, these provisions clearly preclude recovery in the present case.

We first examine the definition of "dishonest or fraudulent acts." The standard form broker's blanket bond is four pages long but contains nine pages of riders. Clause (A) of the standard form provides coverage for "[l]oss through any dishonest or fraudulent act of any of the Employees, ... including loss of Property." Although the standard form expressly defines "Employees" and "Property," it does not expressly define "dishonest or fraudulent act." The first rider converts the standard form, a "Form 14 Broad," to a "Form 14 Basic." In a substitute Clause (A) under the heading "Dishonesty," the first rider eliminates the word "fraudulent," but the rider includes no express definition of "dishonest act."

The second rider, entitled "Definition of Dishonesty--Exclusions Rider for Use with All Financial Institution Blanket Bond Forms (Except Form D) that Do Not Have a Definition of Dishonesty and To Add Certain Exclusions," purports to delete "Insuring Agreement (A)" (presumably the modified Clause (A) from the first rider) and to replace it with a slightly different provision covering "[l]oss resulting directly from one or more dishonest or fraudulent acts of an Employee." Central to the dispute in this matter is the following provision of the second rider:

Dishonest or fraudulent acts as used in this Insuring Agreement shall mean only dishonest or fraudulent acts committed by such Employee with the manifest intent:

(a) to cause the Insured to sustain such loss; and (b) to obtain financial benefit for the Employee, or for any other person or organization intended by the Employee to receive such benefit, other than salaries, commissions, fees, bonuses, promotions, awards, profit sharing, pensions or other employee benefits earned in the normal course of employment.

INA requested the magistrate to give an instruction based on the "manifest intent" clause. He declined, stating that it was up to the jury to read the bond, and told the jury:

[I]f ... you find that the statutory violations were due to fraudulent or dishonest conduct, then the losses are included in the coverage of the bond....

....

If you render a verdict in favor of the plaintiff against MSI and that verdict is based on the fraudulent and dishonest acts of Michael Starbuck or any other MSI employee, then I direct you to find that your verdict will result in a loss to MSI that may be covered under the INA bond.

In denying INA's motion for judgment notwithstanding the verdict, the magistrate amplified his reasoning and opined that the standard form bond contains a definition of dishonesty that renders the second rider and its "manifest intent" language inapplicable. We disagree.

It is true that the second rider applies only to blanket bond forms "that Do Not Have a Definition of Dishonesty." Contrary to the magistrate's conclusion, however, neither the standard form nor the first rider contains such a definition. The standard form merely states that it covers "[l]oss through any dishonest or fraudulent act of any of the Employees, committed anywhere and whether committed directly or by collusion with others, including loss of Property through any such act of any of the Employees." This language explicitly indicates that certain circumstances surrounding dishonest or fraudulent acts do not negate the insurer's obligations. In no sense does it delineate, or purport to delineate, the character of acts falling within the bond's coverage. It thus can hardly be viewed as a definition.

The first rider states that the policy covers "[a]ny loss through any dishonest act of any of the Employees, committed anywhere and whether committed directly or by collusion with others, including loss of Property through any such act of any of the Employees." Again, this statement cannot be reasonably construed as a definition of dishonest or fraudulent acts. The second rider, which contains the manifest intent language, therefore applies.

To trigger coverage the second rider requires that an employee have a "manifest intent" to cause both a loss to the insured and a financial benefit to him or herself. The magistrate's failure to give the requested instruction therefore was error. We need not remand for a new trial, however, because the evidence is legally insufficient to support a finding of the requisite manifest intent. Indeed, as to the first element--manifest intent to cause a loss--the evidence indicated that Starbuck intended to benefit MSIA, no matter how reckless and imprudent his conduct may have been. As to the second element--manifest intent to obtain financial benefit--there was no evidence that Starbuck ever misappropriated any of the funds for his own benefit or even received such funds except for salaries or commissions exempted from the definition by the rider. Thus, the evidence is insufficient to support the requisite finding of manifest intent.

This conclusion also follows from our recent decision in Leucadia, Inc. v. Reliance Ins. Co., 864 F.2d 964, 972 (2d Cir.1988), cert. denied, --- U.S. ----, 109 S. Ct. 3160, 104 L. Ed. 2d 1023 (1989). Leucadia involved a fidelity bond with provisions similar to those in the instant case. We held that such provisions did not insure against losses resulting from an employee's authorizing loans to persons with obvious credit problems and misrepresenting to his superiors the value of the collateral received. Because the employee misguidedly hoped to benefit his employer and received no personal gain from the transaction, we held that the requisite manifest intent had not been shown. See id. at 972-74.

Reversal is also compelled by the bond's provision concerning trading losses. That provision excludes from coverage "[a]ny loss resulting directly or indirectly from trading, including all transactions involving the purchase, sale or exchange of securities, with or without the knowledge of the Insured, ... and notwithstanding any act or omission on the part of any Employee in connection with any account relating to such trading, indebtedness, or balance." In holding that the jury might permissibly find that the losses in question were not within this exclusion, the magistrate relied upon our decision in Index Fund, Inc. v. Insurance Co. of North America, 580 F.2d 1158 (2d Cir.1978), cert. denied, 440 U.S. 912, 99 S. Ct. 1226, 59 L. Ed. 2d 461 (1979), which construed language identical to that contained in the bond in the instant case. Index Fund is distinguishable, however. In that case, the trades were made at artificially inflated prices as a result of bribes. See id. at 1160, 1162. We stated that when "the obligee is a regulated investment company, rather than a broker, fraudulent purchase of securities for the company by the covered employee at a manipulated price may well be considered outside the contemplated meaning of 'trading.' " Id. at 1162.

The obvious purpose of the trading exclusion is to exempt from coverage losses caused by market forces, misjudgments of those forces by buyers and sellers of securities, or various errors or omissions--e.g., a failure to execute an order--in the course of trading. Index Fund, therefore, is easily distinguishable. The loss there was caused by dishonest trading induced by a bribe, and, contrary to the view expressed by the magistrate, some of the loss could not be avoided by a sale while the stock was up. No matter how high the stock might go, the loss incurred in paying an artificially inflated price was of course permanent. Moreover, because of the bribe, the transaction was not one governed by a good faith judgment as to market movements, however poorly founded or even reckless that judgment might be.2 In the present case, it is uncontroverted that MSIA's losses were solely the result of bad trades by Starbuck.3

The manifest intent provision and trading loss exclusion limit protection under this bond to losses due to embezzlement or embezzlement-like acts. The bond thus does not purport to insure against losses caused by reckless or improvident trading, even where the investors have been led to believe that a more conservative investment strategy would be followed. At oral argument, a variety of arguments were made inviting us to consider the financial plight of the plaintiffs in resolving this dispute. We are not indifferent to their plight, but we cannot ignore the fact that insurance systems rely upon predictability of risk and that predictability will be seriously diminished or destroyed if courts refuse to abide by limitations on insurance policies. See Romano, Directors' and Officers' Liability Insurance: What Went Wrong?, in New Directions in Liability Law 77-80 (W. Olson ed. 1988) (37 Proc.Acad.Pol.Sci. 67) (crisis in directors' and officers' liability policies due in part to uncertainty created by "lawless" judicial decisions expanding coverage).4

A refusal to recognize the intended limitation on coverage in the present case would amount to a legal rule that investment firms may not purchase insurance against embezzlement or like acts without also purchasing insurance against an indeterminate number of other risks, including deception by a trader like Starbuck who pursues reckless trading, to be identified only after the fact by courts. Such a rule would necessarily entail extra costs to be borne ultimately by investors. These costs may be substantial because the other risks selected by the courts may be entirely different from embezzlement-like acts. For example, the moral hazard created by insurance covering losses incurred through risky trading such as Starbuck's seems much greater than the moral hazard created by insurance covering embezzlement.5 Even if we were willing to ignore the clear language of the bond, therefore, recovery in this case would be at a cost to future investors, whether in the form of higher premiums, lower limits on insured losses or even the lack of insurance protection against embezzlement.

Reversed.