[Appeared: Communications of the ACM, Dec. '94, V37 N12, pp.13-15.]
See also comments.

Can Wiretaps Remain Cost Effective?

by Robin Hanson

Until recently, technology has happened to allow for cheap wiretaps. New digital telephone technologies, however, may soon make wiretaps more difficult, and new encryption technologies may soon make them almost impossible. This may be good news to privacy buffs, but it worries U.S. police agencies -- since 1968 the law has explicitly allowed police wiretaps. And it worries U.S. spy agencies -- since 1978 the law has explicitly allowed them to wiretap foreigners.

So in early-1992, the FBI proposed a bill to require all telecommunication and computer companies to do whatever it takes to preserve cheap police wiretaps. This bill evolved, became somewhat weaker, and acquired a provision to pay phone companies up to $500 million for costs incurred. In late-1994 it passed.

Also, in April 1993 the White House announced the Clipper chip, a police-wiretap-friendly encryption chip developed by the NSA spy agency, with the hope it will become a de facto standard, at the expense of the total privacy chips on the verge of being marketed by others. If they succeed, they would raise the costs of, but not prohibit, totally private phone conversations. To push Clipper, the Clinton administration has promised to buy lots of Clipper phones and has imposed export restrictions on the alternatives. So far, the result has been to inhibit a private phone market -- neither Clipper nor the alternatives are selling well.

Do the benefits of wiretaps out weigh the many costs of trying to preserve wiretap abilities in the face of technological change? This question isn't easy, but it's not as hard as it looks. We can use some basic economics of law enforcement to help estimate the benefits, and we may be able to solve the problem without answering this question directly.

Let's first consider the costs. Clipper chips would need more power and bandwidth, and would cost more to make than total privacy chips -- about $10 a chip [11] vs. $7 a chip [3] in lots of 100,000. If all 141 million U.S. phone access lines in 1991 [12] had one Clipper phone, this extra chip cost alone would be $424 million. Of course not all phones would be so equipped any time soon, but the wiretap benefit is per phone, so this is the number for comparing costs and benefits.

It will also cost phone companies a great deal to support taps of digital phone conversations, and to follow them through call-forwarding, and so forth. The FBI once claimed $300 million to be the maximum cumulative development cost "for a switch based software solution" to enable phone companies to continue to support wiretaps [6]. However, the president of the U.S. Telephone Association estimates $1.8 billion "to solve the call-forwarding problem alone" [5], and the phone industry consistently maintains that total costs would be $2.5 to $5 billion just in the first four years [2].

More importantly, government imposed constraints and standards would interfere with the development and use of global telecommunications and encryption markets, and could slow down the current telecommunications revolution, raising costs and limiting the features and products offered. Since U.S. phone companies had total annual revenues of $160 billion in 1991 [12], raising costs by just 1% of this figure would cost $1.6 billion. Moreover, the Software Publishers Association estimates that the potential U.S. share of the foreign market for mass-market software with encryption capabilities, now threatened by export restrictions, could total $3 to $5 billion annually by 1997 [7].

Finally, there are continuing costs of privacy lost, not only from legal wiretaps, but also from the risk of misuse of the government's wiretap "backdoor." For example, industrial spies may gain access by corrupting government officials. If the comfort of knowing their phone conversations were totally private was worth only $1 per year per phone line to people, on average, this value would be $ 141 million annually.

Foreign [8] and domestic governments have had a long history of corrupt wiretaps that threaten the proper working of democratic government. A few decades ago, for example, the FBI gained undue influence by wiretapping both members of Congress and the Supreme Court [4]. Martin Luther King and other civil rights leaders were also tapped [11]. The fact that Clipper's algorithm is secret, and that its "two independent escrow agencies" are the Treasury, which runs several federal police agencies, and NIST, which takes the lead from the NSA spy agency on these matters, hardly inspires confidence on this point.

So on the cost side we have at least several billion dollars, and perhaps even that much per year. Now let's consider the benefits.

Wiretaps mainly help police investigate "victimless" crimes -- about 70% of wiretaps regard drug offenses and 10% regard gambling. How much is this help worth? Consider the following statistics offered [6] in support of the conclusion that "the economic benefits alone" of wiretaps are "billions of dollars per year":

During the period from 1985 to 1991, court-ordered electronic surveillance conducted just by the FBI led to 7,324 convictions, almost $300 million in fines being levied, over $750 million in recoveries, restitutions, and court-ordered forfeitures, and $1.8 billion in prevented potential economics loss [6].

If we add these figures and take into consideration that the FBI reports 56% of police wiretap spending [1], that 6% of taps were electronic bugs not threatened by new technologies, and that U.S. spy agencies report about 55% as many wiretaps as police [11], we might estimate $1.2 billion per year of threatened wiretap benefits. At this benefit level, it is plausible, though not obvious, that wiretaps are worth preserving.

This figure underestimates benefits by neglecting noneconomic crime losses. But it also overestimates benefits by neglecting nonwiretap contributions toward achieving these convictions, by assuming these convictions would have been impossible without wiretaps, by simply accepting the FBI's own undocumented estimates of the benefits the FBI provides and by assuming a dollar of fines or of theft prevented equals a dollar of benefit (which it just doesn't [9]). Overall, this estimate is seriously flawed.

We can do better by understanding some long- and well-established results in the economics of law enforcement [9]. The fact is we now delegate most of our policing decisions to public police agencies. That is, our political representatives don't make every little decision about what crime leads to investigate, or what kinds of and how many guns, cars, or desks police should buy. Instead, our representatives set a budget for each police agency, and then focus on broad criteria for evaluating police success. Such criteria include rates of arrests, convictions, reported crimes, and citizen complaints.

Assume a police agency has the correct total budget, and their evaluation criteria induce the correct incentives, to monitor for crime where it would help the most, and to get as many convictions as possible. Then the social benefit of a gun, car, desk, or anything else this police agency buys would be just the amount that agency would have been willing to pay for that item. And the social value of all the cars this agency buys, for example, would be the sum of these willingness values, minus the total price paid for those cars.

Yes, police budgets and evaluation criteria are not exactly optimal. But we don't know which way the bias is -- if we knew that, we would do better to focus on changing the budget and criteria. Thus, the current social value of wiretaps, and hence the amount of additional costs we should be willing to incur to preserve wiretaps, can be reasonably estimated as the total amount police would willingly pay for wiretaps, minus the total they now spend on wiretaps.

In 1993, 976 U.S. police wiretaps were authorized, 17 were never installed, and 55 were electronic bugs, leaving 904 nonbug wiretaps installed. They cost an average of $57,256 each, for a total of $51.7 million spent on legal police wiretaps in 1993 [1]. (I will neglect the possibility that police spent much more than this on illegal taps.) This is less than one part in 600 of the $31.8 billion spent by U.S. police in 1990 [12], and most of it was spent on time the police sat and listened to 1.7 million conversations (at $32 per conversation). So if we again assume 55% more spy spending, we might estimate $80 million per year spent on wiretaps.

So the question becomes: how much more, on average, would police have been willing to pay to make the same wiretaps? One way to answer this would be to conduct a survey of police who make wiretap decisions, asking them how much more they would pay for particular taps. But the FBI will not allow such a survey by civilians, and no state police survey has been attempted, to my knowledge. So we'll have to try and estimate.

Some wiretap installations were no doubt expected to be of marginal benefit (or near-zero value; police would not bother if their costs were just a bit higher). And within even the most valuable installation, many conversations tapped were no doubt also of marginal expected benefit. But other wiretaps were no doubt of higher expected value; police would have paid several times, perhaps even 10 times, what they paid to conduct those taps. So what is the average?

If police would willingly spend, on average, twice as much as they do on wiretaps, the yearly social value of wiretaps would be $80 million. But police would have to be willing to spend 17 times the amount they now spend (or $540 per conversation) in order for this value to reach the $1.2 billion figure estimate based on fines and FBI benefit claims!

Now there are some goods, like water, for which people would willingly pay 17 times what they now pay. But most goods have many similar substitutes, and so people aren't willing to pay very much more than they do. Thus, police wouldn't pay that much more for handguns if rifles were available, nor would they pay that much more for cars if vans were available.

So now this is the question: how good are the substitutes for wiretaps? If unable to wiretap a particular suspect's phone line, police might instead use hidden microphones, informants, undercover agents, stake-outs, go through a suspect's trash, grant immunity to related suspects, or investigate a suspect in many other ways. Or police might choose to focus on other suspects who are more easily investigated without wiretaps. We might also compensate for costlier wiretaps by changing crime punishment levels, perhaps even adding extra punishments for crimes committed with the aid of strong cryptography.

Police are price-sensitive regarding wiretaps -- most cases where a wiretap might help don't use one mainly be cause of their price. Wiretaps cost an average of $14,300 per arrest aided in 1993 [1], or almost five times the $3,000 average spent per non-traffic arrest by U.S. police in 1991 [12].

While these considerations suggest there are good substitutes for wire taps, the FBI and others [6] claim wiretaps are "essential," citing the fact that current law requires police requesting a wiretap authorization to convince a judge that other approaches "reasonably appear to be unlikely to succeed if tried or to be too dangerous."

But judges don't often question boiler-plate claims to this effect in police requests [10]. In 1993 no judge denied any police wiretap request. Also, investigations often continue even after a wiretap has failed to aid an investigation -- yet why bother if the wiretap was "essential"? And how could wiretaps be so essential if nine states felt safe forbidding police to use wiretaps, and if 29 other states didn't bother to use them in 1993? (In fact, 73% of state wiretaps were in just three states: New York, New Jersey, and Florida.)

Given all this, it seems hard to imagine that police would now willingly pay, on average, six times what they pay for wiretaps. Yet even then the wiretap benefit would be only $400 million annually. This figure is high enough to make it possible that wiretaps are worth preserving, but the case would be hard to make. Phone costs would only have to rise by one part in 400 of current revenues to by themselves wipe out such a benefit. Thus, it seems wiretaps are likely not worth preserving -- the age of the economical wiretap is ending.

There is room for disagreement, however. If consensus cannot be obtained on an analysis like mine, and if policy-makers still lean toward preserving wiretaps, then we might do better to rely on improving police incentives, and then delegating such decisions to police.

Consider Pasadena, Calif., which allows its police to rely in part on helicopters to patrol the city. A helicopter is more expensive than a patrol car, more invasive of privacy, and it imposes extra costs on citizens just by being really noisy. But it can help a lot sometimes -- much like a wiretap.

Imagine that the cost of making helicopters was about to double. Would it then make sense for the federal government to require manufacturers to sell helicopters to police at the old prices? What if manufacturers were reimbursed for costs incurred, so small towns could subsidize big city copters? No, it would make more sense to just let prices rise, and to perhaps raise police budgets a bit. Then each local police agency would likely keep their copters only if copters were still worth the cost. Rather than subsidizing police copters, a city might do better to tax copters, forcing police to take into account the extra costs they impose on citizens.

Similarly, if phone companies are experiencing new costs in supporting wiretaps, let them choose wiretap fees and charge local police per wiretap. This makes more sense than a federal wiretap subsidy. If phone companies can't find a profitable price at which local police are willing to buy wire taps, then wiretaps both won't happen and likely aren't worth the cost. And since existing law already requires police to compensate phone companies for "expenses" to assist a tap, this is hardly a radical proposal.

Of course an extra wiretap tax, or some related mechanism, might be required to get local police to take into account nonphone company costs of preserving wiretaps, such as the extra Clipper chip costs and costs imposed by export restrictions. But requiring each police agency to pay, out of their own budget, free-market prices for phone company wiretap support seems at least a minimum requirement.

As with helicopters, the more we can structure police incentives to take into account the full array of costs and benefits of police actions, the better off politicians would be to leave such decisions to the cop on the beat.


1. Administrative Office of U.S. Courts.  Report on applications for
   orders authorizing or approving the interception of wire, oral, or
   electronic communications. Washington, DC, 1993.
2. Berman, J. Testimony before House Subcommittee on 
   telecommunications and finance, Sept. 13, 1994; 
3. Bryen, S. Private communication with Robin Hanson, Aug. 1994.
4. Charns, A. Cloak and gavel, FBI Wiretaps, Bugs, Informers, and The 
   Supreme Court. University of Illinois Press, Chicago, 1992.
5. Charttrand, S. More subsidies sought to pay for wiretap plan. 
   New York Times, p. C2, Aug. 12, 1994.
6. Denning, D. To tap or not to tap. Comm. of ACM 36, 3 (Mar. 1993), 25-44.
7. General Accounting Office. Report on Communications privacy,
   federal policy and actions, to House Committee on the Judiciary. 
   Nov. 1993; ftp://cu.nih.gov/gao-reports/
8. Headrick, D.R. The Invisible Weapon. Oxford University Press, 1991.
9. Posner, R. Economic Analysis of Law, 4th Ed. Little, Brown, and 
   Company, N.Y., 1992:
10. Report of the National Commission for the review of federal and
   state laws relating to wiretapping and electronic surveillance. 
   Washington, DC, 1976.
11. Special Panel of the ACM U.S. Public Policy Committee. Codes,
   keys, and conflicts: Issues in U.S. crypto policy. June 1994; 
12. Statistical Abstract of the United States, 1993.
known by AltaVista
known by AltaVista