The FCC is selling county-sized overlay licenses in the 2.5 GHz (2496–2690 MHz) band in this auction. In each county, up to three categories of licenses will be made available. Category C1 licenses total 49.5 MHz in size, currently consisting of nine 5.5 MHz channels. Category C2 licenses total 50.5 MHz in size, currently consisting of twenty channels between 0.33 MHz, 5 MHz and 6 MHz in size. Category C3 licenses total 17.5 MHz in size, consisting of three 0.33 MHz-channels and three 5 MHz-channels. The 5 MHz-channels are discontiguous from the 0.33 MHz-channels in these Category C3 licenses.

The FCC is selling these county-sized overlay licenses in each category unless the current incumbents cover 100% of the area of the county. The FCC considers these licenses as "fully-encumbered" and not for sale. Roughly 72% of the U.S. population will have at least one overlay license available for sale in this auction. However, as overlay licenses, winning bidders may not get full access to these spectrum licenses. Because incumbents still have a right to operate in these bands, the winning bidders will either have to avoid and protect incumbent operations or negotiate for their removal. This is why this auction is sometimes called a "white space" auction and the FCC calls these "encumbered" licenses.

The FCC has provided very nice maps of the current incumbents in each category of the 2.5 GHz band and the number of channels encumbered. We show them below, but interactive versions these maps and helpful legends can be found on the FCC's Auction 108 Mapping Tool.

At the start of the auction, we told you that we took FCC shape files to calculate the level of encumbrance that each license faces on a population basis. In particular, we intersected the "collapsed" FCC shape files of the licenses and tribal applications provided in the Auction 108 Mapping Tool with 2 x 2 km cell population data (2010 Census)⁽¹³⁾ to determine the population-weighted average number of channels in each county license that would be encumbered. We used the population-weighted number of channels encumbered to classify licenses in each category into "encumbrance classifications" that we can then use in more granular pricing analysis.

But as we noted previously, one cannot easily translate these encumbrance classifications into revised price per MHz-POP figures. This is because the FCC shapefiles provided information regarding the number of channels encumbered inside each license, but not which channel. This is a problem because each channel in the C2 and C3 blocks are not the same size. Nevertheless, the encumbrance analysis provided to date clearly shows that bidders are preferring to pay more for licenses that face less encumbrances...not surprisingly. A few subscribers, however, continue to ask me to provide prices based on just the fully unencumbered MHz-POPs (or "white spaces") inside each license.

Well, those subscribers are in luck.

AllNet Insights and Analytics, who has a great spectrum dashboard tool and who has been tracking the 2.5 GHz band for quite a few years, has graciously provided its estimates of the number of MHz-POPs the are unencumbered for each license...saving me at least a weekend of work. With this data, we are now able to provide pricing analysis for each license based on the unencumbered MHz-POPs. We have inserted this pricing analysis throughout the site with the label "White Space" or "WS" Prices.

I have also subsequently added white space estimates from T-Mobile, which were submitted in the proceeding. This was done after some subscribers indicated that T-Mobile's white space estimates were also valuable as a secondary data point. T-Mobile submitted its estimated number of unencumbered MHz-POPs in each license using 2020 Census data. I converted these numbers into percentages which I then applied to both my 2010 and 2020 numbers so that I can have a complete set of data. I have indicated in footnotes throughout the site which data set was used for my white space analysis.

As with the prior encumbrance classification analysis, several important caveats are necessary:

(1) To be conservative, AllNet's white space percentages assume that every tribal application will be granted regardless of current status. I believe this to be the case with T-Mobile's data as well.

(2) AllNet's white space percentages are estimates and people should do their own analysis. For example, with the intersection analysis that we did with the FCC-provided shape files for the C1 licenses (which are all the same size), our nationwide "white space" price for C1 licenses currently varies from the prices we get using AllNet's percentages by just under 10%. I suspect that some of the difference is due to (a) the fact we are using 2x2 km cell sizes to intersect with the shape files rather than something more granular,(b) perhaps some difference in the licenses which were included or the shape of these licenses, and (c) rounding. Nevertheless, we believe AllNet's percentages are probably best independent estimates available right now.

(3) T-Mobile's white space percentages are also estimates and people should do their own analysis. For example, I noticed that although T-Mobile said that they were using 2020 Census estimates, some counties had more MHz-POPs available then would be possible based on those numbers. I had to cap those white space percentages at 100% under my methodolgy. In addition, T-Mobile's data did not include Alaska or the US Pacific Territories (e.g., Guam, American Samoa).

(4) These "White Space Prices" assume zero value to encumbered MHz-POPs despite this auction making available overlay licenses over the entire county. As bidders are buying licenses even when the MHz-POPs are nearly 100% encumbered (and in some cases 100% encumbered), this "white space" approach has some obvious limitations. For example, the price per MHz-POP for some licenses that are heavily encumbered will be skewed to very large values. I have noted this caveat in several places on the website, but it is worth mentioning again.

To prevent confusion, I have removed the prior encumbrance classification map and replaced with a map that shows the white space availability percentages using AllNet's data immediately below. I have also included these percentages in the table that visitors can download at the bottom of the page. However, for continuity, I will leave a few charts that used the prior encumbrance classifications as well.

Category	Min % of Channels Encumbered (weighted by POPs)	Max % of Channels Encumbered (weighted by POPs)	Encumbrance Classification
Category C1	0%	0%	C1 - 0
Category C1	> 0%	< 50% (or 4.5 channels)	C1 - 50
Category C1	>= 50% (or 4.5 channels)	< 75% (or 6.75 channels)	C1 - 75
Category C1	>= 75% (or 6.75 channels)	<= 100% (or 9.0 channels)	C1 - 100
Category C2	0%	0%	C2 - 0
Category C2	> 0%	< 50% (or 10.0 channels)	C2 - 50
Category C2	>= 50% (or 10.0 channels)	< 75% (or 15.0 channels)	C2 - 75
Category C2	>= 75% (or 15.0 channels)	<= 100% (or 20.0 channels)	C2 - 100
Category C3	0%	0%	C3 - 0
Category C3	> 0%	< 50% (or 3.0 channels)	C3 - 50
Category C3	>= 50% (or 3.0 channels)	< 75% (or 4.5 channels)	C3 - 75
Category C3	>= 75% (or 4.5 channels)	<= 100% (or 6.0 channels)	C3 - 100

Encumbrance Classification	Gross Proceeds	Price MHz POP
C1 - 0	$29,519,910	$0.080568
C1 - 50	$58,431,350	$0.037840
C1 - 75	$24,471,240	$0.017597
C1 - 100	$45,740,510	$0.007569
C2 - 0	$37,434,820	$0.112279
C2 - 50	$75,171,930	$0.051244
C2 - 75	$32,198,170	$0.024818
C2 - 100	$59,165,380	$0.008044
C3 - 0	$21,062,890	$0.066973
C3 - 50	$24,471,230	$0.047005
C3 - 75	$5,246,990	$0.028090
C3 - 100	$14,875,250	$0.007233

The chart below tracks auction progress (using the metric of excess demand as a percent of aggregate demand) against other recent prior FCC auctions.

NOTE: Any comparison between Auction 97 (the AWS-3 auction and an SMRA auction) and the other auctions shown (which are all clock auctions) needs several important caveats. SMRAs have a fundamentally different structures from clocks auctions (e.g., SMRAs allow bidders to bid for specific licenses whereas in clock auctions you bid for generic blocks; SMRAs determine provisional winners after reach round whereas clock auctions just determine aggregate demand at the end of each round). Participants in an SMRA also face different rules regarding the exiting of accepted bid (e.g., a provisionally winning bidder in an SMRA is allowed to submit a withdrawal bid subject to a withdrawal penalty whereas in a clock auction a bidder may not withdraw a bid once demand equals supply). Finally, in SMRAs, bidders do not need to affirmatively bid in every round for licenses that they have provisionally won as they will have an opportunity to bid again in the next round if their provisionally winning bid is outbid in the current round. These differences make any direct comparisons difficult. Nevertheless, with all these caveats in mind and using a simple moving average to “smooth” the variations in demand that you see from round-to-round in an SMRA, I thought that it might still be interesting to make a comparison between Auction 97 and this auction.

In the bubble graph below, we attempt to track how current prices of the 100 largest counties mapped into their corresponding PEAs (on a price per MHz pop basis) compare to their "Predicted Price" based on past FCC auctions. To accomplish this comparison, we need to introduce the concept of the "Overprice Ratio". This Overprice Ratio is the current price of each county's PEA divided by the Predicted Price for that PEA.

This Predicted Price for any county license is calculated by first identifying its relevant PEA and then classifying each PEA into its relevant REAG (an FCC licensing schema that divides the country into 6 regions) and then identifying a "Reference Market" for that REAG. Typically, the Reference Market for each REAG is the largest PEA by population (with the exception of the Southeast REAG where we selected Atlanta as its Reference Market rather than Washington DC, which we have reclassified as being in the Northeast REAG). For those PEAs that are Reference Markets, we use the New York PEA (the "Nationwide Reference Market") as their Reference Market. We then take the current price for each Reference Market and multiply it by the historical ratio of auction prices between each PEA and its Reference Market to get the Predicted Price.

The y-axis on the left shows the log (base 10) of the "Overprice Ratio." We take the log of the Overprice Ratio to lessen the impact of outliers. By definition, the New York PEA (shown in the graph as the biggest bubble in dark blue below), as the Nationwide Reference Market, will always remain at y=0 (the horizontal green line in the graph below). The y-axis on the right side of the graph tracks the current price for New York PEA (based on all the counties that fall into it) after the latest round and is adjusted so that New York PEA's price always remains on the green horizontal line in the graph below.

The x-axis shows aggregate demand in each county (regardless of category) after the latest round. There is a vertical green line where aggregate demand equals 3. This the maximum number of licenses available for sale in each county. Thus, if the bubble shows up to the right of this green vertical line, aggregate demand in this county is greater than supply. The bubbles representing the counties are sized based on their population and color-coded based on the PEA that they fall in.

Example:
Pick San Diego County. It falls into the San Diego PEA (which only has one county in the PEA ). Its Reference Market is Los Angeles.

Suppose
San Diego County current aggregate demand = 10
San Diego current price (average across categories) = $0.50
Los Angeles current price (average across categories) = $0.75
Historical ratio (San Diego/Los Angeles) = 1.09784

Compute
San Diego current Predicted Price = $0.75 * 1.09784 = $0.82338
San Diego Overprice Ratio = $0.50 / $0.82338 = 0.607253
San Diego logarithmic-scaled Overprice Ratio = Log10(0.607253) = -0.21663

Interpretation
In the logarithmic-scaled chart below, San Diego would be represented by a bubble with coordinates (x,y) = (10,-0.21663). That means San Diego is underpriced. All bubbles with negative values are underpriced; conversely, all bubbles with positive values are overpriced with respect to their Reference Market's current price. Before the Reference Market's price converges, any underpriced PEA is a valid prediction due to the non-decreasing price property; however, the converse may not be true until the Reference Market's price converges.

To get the complete list of Reference Markets, Predicted Prices and Overprice Ratios for any county see the table at the bottom of this web page.

⁽¹⁴⁾ Because these deals are based on historical comparisons at the PEA-Level, some caution is necessary. For example, a county that sits at the edge of a large PEA may look underpriced when comparing to the predicted price for the larger PEA in which it resides. Also note that we are not using "white space" pricing analysis here.

County	County Name	FIPS Code	Population	Supply	Aggregate Demand	Gross Proceeds	County Price MHz Pop	PEA	PEA Name	PEA Price MHz Pop	Reference PEA	Reference PEA Name	PEA Predicted Price	Overprice Ratio (Log Base 10)
TN-037	Davidson	47037	626,681	2	2	$620,000	$0.009893	PEA032	Nashville, TN	$0.010578	PEA030	Kansas City, MO	$0.067278	-0.803476
TX-121	Denton	48121	662,614	3	3	$358,000	$0.004598	PEA008	Dallas, TX	$0.005329	PEA001	New York, NY	$0.010061	-0.275984
TX-085	Collin	48085	782,341	2	2	$400,000	$0.005113	PEA008	Dallas, TX	$0.005329	PEA001	New York, NY	$0.010061	-0.275984
CA-111	Ventura	06111	823,318	3	3	$668,000	$0.006905	PEA002	Los Angeles, CA	$0.005944	PEA001	New York, NY	$0.009428	-0.200332
CA-029	Kern	06029	839,631	3	3	$706,000	$0.007156	PEA002	Los Angeles, CA	$0.005944	PEA001	New York, NY	$0.009428	-0.200332
CA-037	Los Angeles	06037	9,818,605	3	3	$6,700,000	$0.005807	PEA002	Los Angeles, CA	$0.005944	PEA001	New York, NY	$0.009428	-0.200332
CA-065	Riverside	06065	2,189,641	3	2	$800,000	$0.005373	PEA002	Los Angeles, CA	$0.005944	PEA001	New York, NY	$0.009428	-0.200332
CA-071	San Bernardino	06071	2,035,210	3	3	$1,400,000	$0.005854	PEA002	Los Angeles, CA	$0.005944	PEA001	New York, NY	$0.009428	-0.200332
FL-086	Miami-Dade	12086	2,496,435	3	2	$994,000	$0.005855	PEA009	Miami, FL	$0.005753	PEA011	Atlanta, GA	$0.008481	-0.168508
FL-099	Palm Beach	12099	1,320,134	3	3	$700,000	$0.004513	PEA009	Miami, FL	$0.005753	PEA011	Atlanta, GA	$0.008481	-0.168508
NC-183	Wake	37183	900,993	2	2	$400,000	$0.004440	PEA045	Raleigh, NC	$0.004881	PEA011	Atlanta, GA	$0.007071	-0.160982
AL-073	Jefferson	01073	658,466	3	3	$260,000	$0.003360	PEA040	Birmingham, AL	$0.004911	PEA011	Atlanta, GA	$0.007017	-0.155001
OH-061	Hamilton	39061	802,374	2	2	$400,000	$0.004985	PEA025	Cincinnati, OH	$0.005619	PEA003	Chicago, IL	$0.007458	-0.123010
FL-095	Orange	12095	1,145,956	2	2	$600,000	$0.005236	PEA013	Orlando, FL	$0.004758	PEA011	Atlanta, GA	$0.005828	-0.088081
FL-105	Polk	12105	602,095	2	2	$200,000	$0.003322	PEA013	Orlando, FL	$0.004758	PEA011	Atlanta, GA	$0.005828	-0.088081
FL-009	Brevard	12009	543,376	2	2	$200,000	$0.003681	PEA013	Orlando, FL	$0.004758	PEA011	Atlanta, GA	$0.005828	-0.088081
AZ-013	Maricopa	04013	3,817,117	3	3	$2,500,000	$0.005574	PEA015	Phoenix, AZ	$0.005574	PEA002	Los Angeles, CA	$0.006681	-0.078666
NV-003	Clark	32003	1,951,269	3	3	$1,100,000	$0.004798	PEA026	Las Vegas, NV	$0.005131	PEA002	Los Angeles, CA	$0.005903	-0.060932
CA-073	San Diego	06073	3,095,313	3	1	$300,000	$0.005538	PEA018	San Diego, CA	$0.005538	PEA002	Los Angeles, CA	$0.006211	-0.049750
GA-121	Fulton	13121	920,581	3	3	$510,000	$0.004715	PEA011	Atlanta, GA	$0.006975	PEA001	New York, NY	$0.007787	-0.047831