Wednesday, May 16, 2012

Markets for adjusting interference rights

One of the problems I tried to solve with the proposal “Stamps and Stewards: A third way to regulate radio operation” was negotiating adjustments to boundaries (e.g. power levels) between unlicensed bands and their neighboring bands given of the collective action challenges faced by unlicensed operators. There are other possibilities; this post explores using auction mechanisms.

Mark Bykowsky and Bill Sharkey have proposed auction mechanisms to solve this kind of problem (Bykowsky & Sharkey 2012), though in their treatment the negotiating parties are all license holders, and even with just two licensees on one side of the deal and single one the other, they note that there may be collective action problems. However, Bykowsky, Olson and Sharkey (2010) suggested an approach for enabling unlicensed operators to participate in an auction. Without taking a position on the merits of either paper, it’s intriguing to explore where a combination of these two approaches might lead. The ideas that follow are based on an illuminating phone conversation with Mark Bykowsky.

Martin Weiss and Liu Cui (2012) have explored spectrum trading with interference rights, but their treatment focuses on trading between primary and secondary users within a band, rather than between parties in adjacent bands with effectively equal status.

A Scenario with Decoupled Receivers

For the sake of concreteness, imagine a band containing receivers not deployed by the licensee in this band (“decoupled receivers”; a “DR” band) where the neighbor (“N”) would like to create more interference into the receivers, e.g. increase its transmission power.

Let’s assume, following the Bykowsky & Sharkey (2011) approach, that an auction was held for the N band license that determined the amount of interference that N could cause to the decoupled receivers, and that it was set Low; in other words, N didn’t value the opportunity to create additional interference into the DR band very highly, and didn’t bid enough to win the High interference right. In this auction various manufacturers and operators of decoupled receivers bid, and won collectively.  The cost of the winning bid was divided among them on the basis of their individual highest bids (cf. Bykowsky, Olson and Sharkey 2010), and they obtained a collective right of protection against any interference above the Low level.

Times, markets and business models change, though, and let’s imagine that N would now be willing to bid High. If there were only one licensee in the DR band, N could just deal with it directly and offer some amount $X up to $ (High - Low) for the right to increase its interference into the DR band. If this was the economically efficient outcome, i.e. $X was more than the harm DR would suffer from the increased interference, a deal would be struck. But let’s say there two hundred DR co-owners of the collective right; there’s a strong likelihood that one of them will hold out for an unreasonably large slice of $X, scuppering the deal.

The way around this would be to arrange another auction a la Bykowsky & Sharkey (2011), with one additional proviso: only DR’s who participated and paid up in the first auction could participate in the second one. All these DRs bid their individual asking prices to tolerate the additional interference, and N bids $X. If $X exceeds the sum of the asks the auction closes, $X is divided among the DRs proportionally to their asks, and N is allowed to deliver more interference to the DR band.

Complications

Various objections and problems are immediately apparent. What about DR manufacturers and operators that came into the market after the first auction? They aren’t allowed to participate in the second auction, and perhaps would’ve asked much more, and prevented the auction from closing. Why should they suffer the additional interference? It’s unfair (for some value of “unfair”).

This solution also assumes that the regulator has the power to change rights in this way, since I assume that the regulator would change the rules allowing N more transmit power, affecting all DRs whether or not they participated in the auction and the resulting pay-off. IANAL, but my guess is that this approach to rulemaking flies in the face of the U.S. Administrative Procedure Act, and perhaps even common law, and that the FCC wouldn’t be able to contemplate this kind of action without new legislative powers, if it were possible at all.

Even if it were fair and legal, a market solution may not work. The arguments and examples in Bykowsky & Sharkey (2011) assume market participants have complete information, that efficient outcomes will be selected, and that there aren’t coordination problems among the bidders. Theirs is a small numbers treatment; there are only two bidders in the category I’ve called DR (“S-type” in their paper). What happens with hundreds of bidders? While there is literature that suggests that public goods can be provided by large groups (Bonacich, Shure, Kahan & Meeker 1976; Isaac, Walker & Williams 1994), and that cooperation can increase with group size, I’m skeptical that these results transfer directly to this case.

There’s also the problem of double counting harm. The papers I’ve seen in this area assume the group members differ only in their preferences; however, in this case one may run the risk of double counting since bidders on the DR side may come from various parts of the value chain, e.g. manufacturers and users. The user of the decoupled receiver may calculate their ask based on the incremental cost to the device of tolerating increased interference from N; and the manufacturer of the device, also participating in the auction, may do likewise. The aggregate ask would, in this case, exceed the actual harm by counting it twice, and therefore an auction that would’ve closed without double counting, will now fail.

Finally: who would make the market? One assumes regulators like the FCC, since they have always run spectrum auctions to date. However, the interference reconfiguration auction described here has money flowing between radio operators – from N to the set of DR, in this case – and not from licensees to the regulator as in auctions to date. (The upcoming TV band incentive auction marks an evolution towards situation.) The state doesn’t benefit directly from the readjustment of rights, so why should it bear the cost of making the market? On the other hand, the regulator plays an essential role in changing the transmit power rules because the decoupled receivers are an indeterminate group, and not tied to a license. The delegation of adjustments to a secondary market where transactions take place between licensees is not possible here.

Another scenario: Unlicensed devices

A similar case would involve an unlicensed (“UL”) band where a neighbor N seeks less interference, i.e. N would like the UL devices to reduce their transmit power. Let’s leave aside the question of whether Part 15.5 (b) prohibiting unlicensed devices from causing interference applies across band boundaries, and assume that N would not be able to persuade the FCC to make this change purely on the basis of Part 5.15.

Again, N would be willing to pay some amount to all the ULs to persuade them to reduce transmit power, but coordinating the deal is difficult. Just as in the previous case, an auction where all the ULs posted their asks in return for reduced operating power could lead to the efficient outcome, which might be that N paid them off.

Likewise, an auction could work if N wanted to increase interference to UL.  This assumes, however, that that the unlicensed devices wouldn’t just have to live with it under Part 15. There is some basis for protecting unlicensed operations vs. licensed in the FCC’s proposal that licensed Location and Monitoring Service (LMS) devices at 902-928 MHz should not cause “unacceptable levels of interference” to unlicensed devices in that band, see Mitch Lazarus “FCC Seeks Comment on Interference into Unlicensed Devices.

Update: Per Harold Feld's comments below and email to me on June 4, 2012, 47 CFR 90.353  (d) requires that "EA multilateration LMS licenses will be conditioned upon the licensee's ability to demonstrate through actual field tests that their systems do not cause unacceptable levels of interference to 47 CFR part 15 devices." 47 CFR 90.361 specifies that Part 15 operations may not cause harmful interference to LMS systems in the 902.928 MHz band, but provides a list of safe harbor exclusions of operations that will not be considered to be causing harmful interference to an LMS system.

References

Bonacich, P., Shure, G. H., Kahan, J. P., and Meeker, R. J. (1976). Cooperation and group size in the N-Person prisoners' dilemma. Journal of Conflict Resolution, 20(4):687-706.

Bykowsky, M. M., Olson, M., and Sharkey, W. W. (2010). Efficiency gains from using a market approach to spectrum management. Information Economics and Policy, 22(1):73-90.

Bykowsky, M. M. and Sharkey, W. W. (2012). Using a market to obtain the efficient allocation of signal interference rights. FCC Staff Working Paper 4. http://www.fcc.gov/working-papers/using-market-obtain-efficient-allocation-interference-rights

Isaac, R. M., Walker, J. M., and Williams, A. W. (1994). Group size and the voluntary provision of public goods: Experimental evidence utilizing large groups. Journal of Public Economics, 54:137-36.

Weiss, Martin B.H. and Cui, Liu (2012). Spectrum Trading with Interference Rights. In: 7th International Conference on Cognitive Radio Oriented Wireless Networks (CrownCom), June 18–20, 2012, Stockholm, Sweden. (In Press)

2 comments:

Harold Feld said...

Mitch's post fundamentally misconceives the existing M-LMS rules. This isn't a "proposal." It's been a rule since 1995.

Harold Feld said...

Mitch Lazarus doesn't understand the relevant rules here. The rules preventing interference by M-LMS licensees go back to 1995, as do the Part 15 safe harbor for the band.