FCC Econometric Analysis of
Potential Discrimination
Utilization Ratios for Minority-
and Women-Owned Companies
in FCC Wireless Spectrum
Auctions
December 5, 2000
A Study Prepared by
Ernst & Young LLP
for the
Federal Communications Commission
FCC Econometric Analysis of Potential Discrimination
Utilization Ratios for Minority- and Women-Owned Companies in FCC
Wireless Spectrum Auctions
1. Introduction and Executive Summary
Utilization ratios are a means of measuring the participation and success of minority- and
women-owned businesses in the Federal Communications Commission (FCC) wireless
spectrum auctions. In this context, a utilization ratio indicates the share of, say, minority-
owned companies that were successful in obtaining a spectrum license through an auction
out of all minority-owned companies that participated in the auction. Comparison of
these ratios across groups highlights any potential systematic differences in auction
outcomes.
There are different ways of calculating utilization ratios, depending on how success and
participation are measured. Perhaps the simplest way of calculating utilization ratios for
spectrum auctions is to determine the number of auction participants who won at least
one license as a percentage of all participants. While conceptually straight-forward, this
measure does not capture, for example, potential differences in the rates at which
participants qualified to bid , or any differences in the value of the licenses won . In order
to develop a more thorough understanding of auction outcomes for minority- and women-
owned companies, this report presents a number of different utilization ratio measures:
? General Utilization Ratio. Percentage of auction winners (those who won at least one
license) among all auction applicants.
? Qualifying Ratio. Percentage of applicants who qualify to bid among all auction
applicants.
? Success Ratio. Percentage of auction winners (those who won at least one license)
among all qualified auction applicants.
? Economic Value Ratio. An alternative way of assessing the extent to which minority-
and women-owned companies are able to secure wireless spectrum licenses is to
evaluate their share of the total economic value of the licenses auctioned. The
Economic Value Ratio is defined as the economic value of licenses secured by
applicants in a particular group, for example minority-owned companies, expressed as
a percentage of the total economic value of licenses auctioned.
? Average Revenue per Winner. If the number of winners in a group is small, then the
Economic Value Ratio for that group would naturally be low, even if there were no
systematic differences in the values of licenses obtained across groups. Average
revenue per winner provides another measure of auction outcomes for different
groups, while controlling for the number of winners.
? Return on Payment Ratio. Another approach to scaling the revenues generated by
various groups is to use the upfront payments as a scaling factor. The upfront
payments determine bidding eligibility and hence affect auction outcomes. The return
on payment ratio is calculated as the percentage of net FCC revenues relative to the
percentage of upfront payments. For example, suppose that minority-owned
companies generated 15% of net FCC revenues and paid 10% of the upfront payments
in a particular auction. Then the Return on Payment Ratio for minority-owned
companies is 1.5, indicating that the minority share of revenues is 1.5 times the
minority share of upfront payments.
Each of these ratios presents a different view on how minority- and women-owned
companies obtain licenses through the spectrum auction process. In general, these ratios
can be classified into two categories: measures of auction outcomes, and measures of
financial implications of those outcomes.
The general utilization ratio provides an overall view of the auction outcomes, based on
the number of applicants and winners. The qualifying and success ratios refine this
concept further, by breaking the auction process into two parts: qualifying to bid, and
winning after qualifying. Contrasting these two measures will identify whether any
differences found in the general utilization ratios are attributable to differences in
qualifying or in succeeding after qualifying, thus providing insights into where in the
auction process there may be differences across groups of applicants.
The other measures are more financial measures of auction outcomes. Simply counting
winners and losers may obscure the fact that some licenses are more valuable than others,
and that there may be differences in the value of licenses acquired across groups of
applicants. The financial measures of auction outcomes address these issues by
examining revenue shares, and scaling them by numbers of applicants or by upfront
payments.
Both types of measures are useful in evaluating the ability of minority- and women-
owned companies to acquire spectrum licenses through auctions, as they provide different
points of view. Taken together, they will provide a more comprehensive picture of
auction outcomes across different groups of auction applicants.
Our main findings, by ratio type, from this analysis are:
Aggregate Utilization Ratios:
? When participation and success is measured by counting the percentages of winners
from all auction participants (general utilization ratio), minority and women
applicants appear to be somewhat less likely to win at least one license relative to
other applicants. These differences are statistically significant.
? Examination of the qualifying ratio indicates that minority and women applicants tend
to qualify at lower rates than other applicants and that these differences are
statistically significant. On the other hand, analysis of success ratios reveals that
among qualified applicants, there are no statistically significant differences between
women and other applicants in their likelihood of winning licenses. Success ratios
also indicate that on average, qualified minority applicants are more likely to win than
qualified non-minority applicants. This difference is statistically significant. These
findings would suggest that the difference in general utilization ratios may be largely
attributable to the differences in qualifying ratios where minority applicants face a
lower likelihood of qualifying. However, once qualified, minorities appear to have
higher odds of success in auctions.
? Figure 1 presents the differences in average general utilization, qualifying, and
success ratios between minority and non-minority applicants.
? Indicates that differences in the ratios across non minority and minority
groups were statistically significant at the 95% level
Impact of Installment Plans
? When the various utilization ratios are analyzed separately for auctions with and
without installment plans, it appears that installment plans generally increase the rate
at which minority and women applicants win licenses. Figure 2 presents the same
ratios as Figure 1 but calculated separately for auctions with installment plans and
without installment plans. Although the utilization and qualifying ratios are still
lower for minorities than for non-minorities for both auctions with and without
installment plans, in auctions with installment plans the success ratio is higher for
minorities than for non-minorities and the difference is statistically significant.
? Indicates that differences in the ratios across non minority and minority
groups were statistically significant at the 95% level
? The difference in outcomes for minorities between auctions with and without
installment plans may reflect various factors. Installment plans may relax potential
capital constraints, or, alternatively, they may lead to inflation of the price of the
license and aggressive bidding. Further study is required to definitively evaluate the
effects of installment plans.
Economic Value Analysis
? The analysis of economic value shares (i.e., shares of net FCC revenue generated
from the auctions) revealed that in the aggregate, the value of licenses acquired by
minority winners is approximately 12% of the total value of licenses. However, in
auctions with installment plans, the minority share of total value increases to
approximately 19% (see Figure 3). The value shares for women winners exhibit a
similar pattern.
Figure 3
Economic Value Shares
? When the economic value of licenses acquired is examined on a per winner basis
(average economic value per winner), there are no statistically significant differences
between minority and women applicants, and other applicants at the aggregate level.
In other words, while the number of minority and women winners is relatively small,
compared to the number of other winners, the value of their licenses won is
comparable to that of other winners, on average.
? When economic value of licenses relative to the upfront payments is compared across
different applicant groups (return-on-payment ratio), minority and women applicants
tend to obtain a larger share of the economic value of the licenses than their share of
upfront payments. If upfront payments are taken as a measure of the value and
number of licenses applicants are interested in winning, this finding would seem to
indicate that minority and women applicants tend to win at least as many and/or as
valuable licenses as they are interested in winning. However, upfront payments may
be an imperfect indicator of interest, if applicants are unable to make upfront
payments in the amounts they desire. Figure 4 illustrates the average return-on-
payment ratios for minorities across all auctions.
? When return-on-payment ratios are analyzed separately for auctions with and without
installment plans (Figure 5), the ratios for minorities are much higher in auctions with
installment plans. Ratios for women have a similar pattern.
Other findings
? The analysis by industry groups revealed few statistically significant patterns across
industries in the comparison of utilization ratios between minority and women
applicants, and other applicants. In advanced paging/data auctions all three measures
of utilization (general utilization, qualifying, and success ratios) are significantly
lower for minorities than for non-minorities.
? The differences in utilization ratios between minority and women applicants, and
other applicants are typically less pronounced among small companies than among
large companies. Among small companies, auction outcomes are generally more
comparable across applicant groups than among large companies.
? In the first three auctions, minority and women applicants were eligible for bidding
credits. In these auctions, the economic value shares tended to be generally larger for
minority and women applicants than in other auctions.
When interpreting the utilization ratio calculations, it is important to keep in mind that
they are based on a high-level analysis that does not control for many important applicant
characteristics that may affect auction outcomes and may provide explanations for the
observed differences across applicant groups. For example, the findings suggest that
installment plans increase the likelihood of winning for minority applicants, which may
reflect the easing of capital constraints, if any, or inflation in the value of licenses. These
results may also be the artifacts of differing auction strategies employed by different
participants. For example, large companies may strategically place upfront payments
across a wider array of spectrum than their business needs, and focus their interest as the
auction proceeds, leading to a lower return-on-payments. We are currently developing
further more detailed analysis into the determinants of auction outcomes.
The remaining sections of this report provide a more detailed discussion of our analysis.
2. Data
The FCC provided Ernst & Young LLP (E&Y) with data on 19 auctions that E&Y used
to calculate preliminary utilization ratios. For each auction applicant, the data include
the following information:
? Indicator for Qualifying Applicant
? Name
? Indicator for Small Business
? Indicator for Rural Business
? Indicator for Woman-owned Business
? Indicator for Minority-owned Business
? Number of High Bids
? Total Population in Areas for each winning bid
? Sum of High Bids $ (Net)
? Sum of High Bids $ (Gross)
? Upfront Payment
? Bidding Eligibility
The indicators for minority-owned, women-owned, and small businesses are based on
self-reported data. It is our understanding that the FCC has not asked applicants to verify
their minority- or women-owned status, but when an auction applicant identified itself as
a small business, it also had to provide some verification to the FCC of its small business
status.
3. Utilization Ratio Calculations
Each of the utilization ratios is calculated separately for each auction, as well as for
groups of auctions. Aggregating data across auctions allows us to detect general high-
level patterns, if any, that might not emerge in an auction-by-auction view. Two natural
groupings of auctions are by rule structure and by industry type. Auction rules have
varied considerably, and in particular, some auctions have provided special installment
plans for small businesses while others have not. An interesting question is whether these
special programs had an impact on participation of minority/women-owned businesses.
Similarly, the licenses have been sold in a number of distinct industries, and there might
be some industry-specific factors influencing the extent of minority/women-owned
business participation in auctions. The groups of auctions are :
? Auction Rule Structures Group
Group 1: Installment Plans (auctions # 2,3,5,6,7,10,11)
Group 2: No Installment Plans (auctions # 1,4,8,9,12,14,15,16,17,18,20,21)
? Industry Groups
Group 1: Advanced Paging/Data (auctions # 1,3,18)
Group 2: Mobile Voice and Data (auctions # 4,5,7,10,11,12,14,15,20,21)
Group 3: Interactive Data (auction # 2)
Group 4: Wireless Cable (auction # 6,17)
Group 5: Multichannel Video (auction # 8,9)
In another grouping, applicants are grouped on the basis of their self-reported size into
small and large companies , and utilization ratios are calculated separately for each
group. Company size may be one of the determinants of auction outcomes, and hence it
is instructive to conduct utilization ratio comparisons for groups of companies that are
similarly sized.
In addition to presenting the utilization ratios for the various groups, the report lists
aggregate (or average) utilization ratios. These are calculated in two ways: simple
averages, and revenue-weighted averages. First, we calculate the aggregate utilization
ratios by using the total numbers of auction applicants and winners across all auctions.
Second, we calculate average utilization ratios from the per-auction ratios that are
weighted by auctions’ revenue shares. For example, suppose that Auction 1 generated
10% of revenues from all auctions combined, and that Auction 2 generated 15% of all
revenues. Then, in the calculation of the average revenue-weighted utilization ratio for,
say, minority applicants, the minority utilization ratio from Auction 1 would receive a
weight of 10%, and the ratio from Auction 2 would receive a weight of 15%. Both
average utilization ratio measures are reported at the bottom of each table for reference.
For each ratio, E&Y conducted tests to determine whether the differences in the ratios
across groups were statistically significant. In these tests, minority applicants are
compared to non-minority applicants (i.e., all other applicants who did not identify
themselves as minority applicants), and women applicants are compared to all other
applicants (who did not identify themselves as women-owned companies). The concept
of statistical significance is used to differentiate between systematic patterns and chance
occurrences. Even in the absence of any systematic differences auction outcomes
between, say, minority and non-minority auction applicants, one would not expect the
utilization ratios for the two groups to be exactly the same. The question is then whether
a difference in the ratios indicates a systematic pattern. Statistical significance tests are a
common approach to determining this. For example, when the statistical significance test
indicates a 95% level of confidence, there is a 95% chance that a difference in utilization
ratios is due to a systematic pattern and only a 5% chance that there is no true difference.
The 95% level is typically used as the threshold level for deeming a result statistically
significant, i.e. if the confidence level is at least 95%, the result is considered statistically
significant; otherwise it is not. The confidence level of a test is affected by not only the
absolute magnitude of the difference in utilization ratios, but also by other factors, such as
the number of applicants on which the utilization ratios are calculated. Hence, a small
difference may at times be statistically significant, while a large one might not.
Although the full population of auction applicants is used in the calculations presented in
the body of this report, tests of statistical significance are a good benchmark for
comparison across groups, given the interest in examining utilization ratios in the auction
process in general, rather than in a specific auction at a specific point in time. Since the
auctions are an on-going process, the populations that participated in each of the auctions
can be viewed as draws from a “superpopulation,” and statistical testing of hypotheses is
appropriate.
A. General Utilization Ratio
The general utilization ratio is an overall measure of the extent to which
minority/women-owned businesses participate in auctions. It is calculated as the
percentage of winners among all auction applicants. Table 1 shows the general utilization
ratios by auction as well as for all auctions as a whole. For all auctions as a whole, there
are some differences across general utilization ratios across demographic groups: 37.40%
of non-minority applicants and 31.97% of minority applicants win licenses, while the
shares are 32.01% for women-owned applicants and 37.35% for other applicants. Both
of these differences are statistically significant, indicating that on average, minority- and
women-owned firms win licenses at slightly lower rates than other firms. The revenue-
weighted average general utilization ratios provide similar information: 28.89% for non-
minorities vs. 22.02% for minorities, and 18.81% for women applicants vs. 29.18% for
other applicants.
In the auction-by-auction comparison, the percentage of winning minorities is sometimes
larger (e.g., auctions 2 and 7) and sometimes smaller (e.g., auctions 6 and 17) than the
corresponding percentage for non-minority applicants. Similarly, women applicants
sometimes win licenses more frequently than other applicants (e.g., auction 17) and
sometimes less frequently (e.g., auction 11). However, the number of minority and
women applicants and/or winners is too small in most auctions (13 out of 19) to allow the
use of statistical tests, so for most auctions, it is unclear what, if any, underlying patterns
these differences represent on a per-auction basis. In the auctions where the number of
minority and women applicants and/or winners permits the calculation of valid tests, none
of the differences are statistically significant.
When auctions are analyzed by auction groups based on auction rules (Table 2),
minorities appear to be just as likely to win licenses in auctions with installment plans,
whereas there is a statistically significant difference between minorities and non-
minorities in auctions without installment plans, with minorities winning licenses less
frequently than non-minorities. In other words, it appears that the presence or absence of
installment plans does affect the outcomes for minorities, as measured by the general
utilization ratio. On the other hand, installment plans appear to make less of a difference
for the outcomes for women auction applicants. Women are less likely than other
applicants to win licenses in auctions with installment plans. This difference is
statistically significant.
The analysis by industry group (Table 3) shows that minority applicants win licenses at
significantly lower rates than non-minority applicants in advanced paging/data auctions
but not in other industry groups. Women applicants also win licenses less frequently than
other applicants in advanced paging/data auctions, as well as in mobile voice and data
auctions.
When auction applicants are grouped by size (Table 4), there are no systematic
differences in outcomes between minority/non-minority and women/other applicants
among small companies. In contrast, both minority- and women-owned applicants have
significantly lower general utilization ratios among large companies.
B. Qualifying Ratio
In order to bid in the auctions, applicants must qualify by submitting Form 175 (“short
form”) and an upfront payment. The upfront payment depends on desired bidding
eligibility. Bidding eligibility is measured in bidding units, and each license to be
auctioned requires a certain number of bidding units. For example, an applicant who
provides an upfront payment that qualifies him for 100 bidding units is able to bid on any
combination of licenses such that the total bidding eligibility requirement does not exceed
the 100 units. Failure to make the upfront payment or to complete the “short form” will
result in disqualification. A potential reason explaining the differences in general
utilization ratio between minority and non-minority applicants and between women and
other applicants, described above, is differences in rates at which minorities and women
qualify to bid, relative to other applicants. To examine the impact of the qualifying
process, we study the qualifying ratio separately from the general utilization ratio.
The qualifying ratio is defined as the percentage of minority/non-minority and
women/other applicants that qualify to bid. Table 5 shows qualifying ratios by auction as
well as for all auctions as a whole. The average qualifying ratios across all auctions
indicate significant differences between the outcomes of minority and women applicants,
relative to other applicants. The average qualifying ratio for minority applicants is
48.3%, while it is 67.5% for non-minority applicants. Similarly, the average qualifying
ratio for women applicants is 51.9%, and 66.6% for other applicants. Both of these
differences are statistically significant, indicating that there are systematic differences in
qualifying rates between minorities and women applicants as compared to other
applicants.
In eight of the 19 auctions, the number of minority/women applicants and/or qualified
applicants is too small to permit the calculation of statistical tests regarding the
significance of the difference in qualifying ratios. In other auctions, the tests indicate
some statistically significant differences in qualifying ratios. Minorities qualify at
significantly lower rates than non-minority applicants in auctions 3, 5, 6, 10, and 11.
Women qualify at significantly lower rates than other applicants in auctions 5, 10, and 11.
Table 6 presents the qualifying ratios when auctions are grouped by auction rules, i.e.,
whether auctions had installment plans or not. Minorities qualify at significantly lower
rates regardless of whether auctions had installment plans or not. Women qualify at
significantly lower rates in auctions with installment plans, but no such difference appears
in auctions without installment plans.
In the analysis by industry group, minorities again qualify at significantly lower rates in
three of the five industry groups (advanced paging/data, mobile voice and data, and
interactive data). The number of minorities is too small to permit statistical tests in
wireless cable auctions, and there are no statistically significant differences in qualifying
ratios in the fifth industry group, multichannel video. For women applicants, the only
industry group with statistically significant differences in qualifying ratios is mobile voice
and data. As for minority applicants, the number of women applicants is too small in
wireless cable auctions to permit the calculation of statistical tests.
Company size appears to make no difference in qualifying ratios. When large
minority/women companies are compared to other large companies, they qualify at
statistically lower rates. The same holds in the comparison of small minority/women
companies to other small companies.
These findings from the study of the qualifying ratio indicate that at least some of the
differences evident in general utilization ratios between minority/women applicants and
other applicants can be attributed to differences in qualifying ratios. In general, minority
and women applicants tend to qualify at significantly lower rates than other applicants. A
possible reason for the lower qualifying rates may be differential access to capital, which
may hinder the ability of minorities and women to make the necessary upfront payments.
However, a more in-depth analysis is required to arrive at authoritative conclusions about
the reasons behind this apparent disparity.
The next step in the analysis is to evaluate success rates for qualified applicants to
determine whether after qualifying there are differences in auction outcomes between
minority and women applicants relative to other applicants.
C. Success Ratio
The success ratio is a measure of qualified applicants who win bids, and is calculated as
the percentage of winners among qualified applicants.
Table 9 shows the success ratios by auction as well as the average success ratios. After
having qualified to bid, minority applicants in fact win licenses at a higher rate than non-
minority applicants (66.1% vs. 55.4%), and this difference is statistically significant.
Qualified women applicants also win licenses at a higher rate than other applicants
(61.7% vs. 56.1%), although this difference is not statistically significant. Again, in 13 of
the 19 auctions, the number of minority and women qualified applicants and winners is
too small to permit the calculation of statistical tests. In auctions where testing is
possible, however, minorities often win at significantly higher rates (auctions 2, 5, and
11). No such systematic pattern appears in the comparison of women applicants and
other applicants on a per auction basis.
When success ratios are examined by auction group, minority and women applicants are
more likely to win than other applicants in auctions with installment plans. However, in
auctions without installment plans minorities win less frequently than non-minorities and
the difference is statistically significant. Women applicants also win less frequently but
the difference is not statistically significant. These results mirror those from the analysis
of general utilization ratio: again, installment plans appear to enhance the ability of
minorities in particular to secure wireless spectrum licenses, while the difference is less
marked for women applicants.
Analysis of auctions by industry groups reveals some differences across industries. While
minority applicants are statistically more likely to win in mobile voice and data auctions
than non-minority applicants, they are less likely to win other industry groups (with the
exception of wireless cable, for which the low number of minority qualified applicants
precludes the use of statistical tests). For women applicants, the only significant
difference emerges in mobile voice and data auctions, in which they are less likely to win
than other applicants.
When applicants are grouped by size, there are no differences in success ratios among
large companies, either between minority and non-minority applicants or between women
and other applicants. However, among small companies, minority applicants win licenses
more frequently than non-minorities, and the difference is statistically significant.
In sum, the success ratios indicate that among qualified applicants, minorities and women
tend to win licenses just as frequently as other applicants, if not more frequently. This is
in marked contrast to the findings from the analysis of qualifying ratios. Those findings
showed that minority applicants in particular tend to qualify less frequently than other
applicants. On the basis of these results, the lower general utilization ratios for minority
and women applicants appear to be related to hurdles in qualifying to bid, but that among
those who qualified, the ability of minority and women applicants to secure licenses is
comparable to that of other applicants.
D. Economic Value Ratio
The preceding measures of utilization (general utilization ratio, qualifying ratio, and
success ratio) have measured utilization as qualifying to bid for or winning at least one
license, relative to the number of applicants. These calculations omit any considerations
of the economic value of the licenses obtained.
An alternative measure of the extent to which minority/women-owned businesses win
FCC licenses is their share of the total economic value of licenses.
Table 13 presents the Economic Value Ratios by auction as well as for all auctions as a
whole. For all auctions as a whole, the value of the licenses won by minority applicants
was 11.9% of the total value of licenses. For women applicants, the share was 7.9%. In
other words, minority (women) applicants secured roughly a tenth of the licenses, when
measured by the economic value of the licenses. In the auction-by-auction analysis, the
value shares for minority applicants have ranged from a low of 0.0% to a high of 31.6%.
The range for women applicants is from 0.0% to 45.2%.
Economic value ratios by auction group are provided in Table 14. In auctions with
installment plans, the economic value ratios for both minority and women applicants are
higher than on average (19.4% and 12.7%, respectively). Consistent with the evidence
from the general utilization ratios and success ratios, these figures indicate that in
auctions with installment plans, minority and women applicants tend to acquire larger
shares of the spectrum for sale, as measured by the economic value of the spectrum.
The economic value shares by industry group are shown in Table 15. In terms of
economic value, minority and women applicants obtain their largest shares of the total
value of spectrum auctioned in multichannel video auctions.
In table 16, economic value shares are calculated by company size. Among small firms,
minority and women applicants obtain higher value shares than on average (22.1% and
14.6%, respectively). In contrast, their shares are very low among large firms (less than
1%).
The findings from the analysis of economic value shares indicates that in general,
minority and women applicants acquire approximately a tenth of the licenses, as
measured by their economic value. However, in auctions with installment plans and
among small firms, their shares of economic value are larger.
E. Economic Value Per Winner
The economic value ratio calculations show that the bulk of the revenues collected from
auctions come from non-minority owned companies, or companies not owned by women.
However, this reflects at least partially the fact that the number of minority and women
winners is smaller than the number of other winners. In order to evaluate the economic
value of licenses acquired by minority and women applicants while controlling for the
number of winners, we calculated the average economic value per winner.
Table 17 displays the results on a by-auction basis, and also gives the averages across all
auctions. On a per winner basis, the differences in economic value of licenses obtained
between minority and women applicants on one hand, and other applicants on the other,
appear to be less distinct than in the preceding analysis of overall value shares. The
average value per winner for minority applicants is $22.5 million, while it is $32.0
million for non-minority applicants. When comparing women applicants to other
applicants, the average values are $15.4 million for women and $33.2 million for others.
Neither of these differences is statistically significant. In the analysis by auction, the
average value of a license per winner is significantly lower for minorities in two auctions
(Auctions 6 and 11). In 10 out of the 19 auctions, there are not enough minority winners
to perform a valid test of statistical significance for the average value of licenses won.
The findings for women applicants are quite similar. In Auctions 6 and 11, there are
statistically significant differences between women and other winners in the average
economic value of licenses won, with women having lower values. In 9 of the 19
auctions, there are not enough women winners for the calculation of statistical tests.
When auctions are grouped by auction rules (Table 18), differences in the average values
of licenses per winner are not statistically significant in auctions with installment plans,
but they are in auctions without installment plans. This finding echoes those from the
previous calculations in which minority and women applicants had higher general
utilization and success ratios, and higher economic value shares in auctions with
installment plans.
In the analysis of average economic value per winner by industry group (Table 19), the
only significant difference is found in the comparison between minority and non-minority
winners in interactive data auctions. No significant differences are found in other
industry groups, or in the comparison between women and other winners. However, there
are no minority or women winners in wireless cable auctions so the tests cannot be
calculated for this industry group.
Table 20 presents the average economic values per winner by company size. Again, the
differences are significant among large companies (i.e., large minority- and women-
owned companies pay on average significantly less for their licenses than other large
companies), but not among small companies. In fact, for minority small companies, the
average economic value per winner is higher than for non-minority small companies.
Taken together, the findings from the analysis of average economic values per winner
suggest that in general, there appear to be no systematic differences in the average values
between minority and women winners, and other companies. While in some auctions the
differences are significant, as they are also among large companies, this is offset by the
fact that in other auctions and among small companies, the differences are small enough
to be statistically insignificant.
F. Return-on-Payment Ratio
The final utilization ratio analyzed in this report is the return-on-payment ratio. The idea
behind this ratio is to scale the economic value of licenses won by the upfront payments.
Upfront payments determine the number of licenses the applicant is able to bid on, and
are indicative of the number and value of licenses the applicant is interested in
acquiring. Therefore we would expect applicants who made larger upfront payments to
win more licenses and/or more valuable licenses. Differences in the value of licenses
won might then be related to differences in upfront payments, and it is interesting to
examine to what extent the value of licenses won, relative to upfront payments, differs
across groups of applicants.
The return-on-payment ratio is calculated as the percentage of net revenues from
minority/women winners relative to the percentage of upfront payments from
minority/women applicants. For example, if minority applicants generated 15% of the
total FCC net revenue, and paid 10% of the total upfront payments, then this ratio would
be 15%/10%=1.5. The calculations of this ratio are limited to 11 of the 19 auctions
because we did not have data on upfront payments for all auctions.
Table 21 presents the overall return-on-payment ratios as well as the ratios by auction.
Overall, the value share of licenses won by minority applicants is nearly twice as large as
the minority share of upfront payments, as indicated by the return-on-payment ratio of
1.86. For non-minority applicants this ratio is close to 1. Similarly, for women
applicants the ratio exceeds 2, whereas for other applicants it is nearly 1. In other words,
minority and women winners tend to generate relatively more revenue than their share of
upfront payments. If upfront payments are taken as a measure of the number and/or value
of licenses applicants are interested in winning, this finding would seem to indicate that
qualified minority and women applicants win at least as many and/or as valuable licenses
as they are interested in winning. Of course, if minority and women applicants are not
able to make as large upfront payments as they would like, upfront payments are an
imperfect indicator of the extent of their interest in licenses. Nevertheless, throughout
this analysis of utilization ratios it is important to keep in mind that differences in the
number and value of licenses won reflect at least partly differences in applicants’ interest
in licenses.
In the analysis by auction, the ratio has been relatively constant and close to 1 for non-
minorities across all auctions, while for minorities the ratio has varied from 0 to nearly 4.
Similarly, while the ratio varies from nearly 0 to almost 2 for women winners, it has been
close to 1 for other winners.
When auctions are examined by auction group (Table 22), again minorities and women
tend to generate relatively more revenue than their share of the upfront payments in
auctions with installment plans. This result again shows that the outcomes for minority
and women applicants appear to be influenced by the existence of installment plans.
Table 23 presents the results from the analysis by industry group. Because of lack of
data, we were unable to calculate the return-on-payment ratio for two of the five industry
groups. In the remaining three groups, the ratios exceed one for minorities in two groups
and are well below one in one group, while they are close to one for non-minorities in all
groups. The ratios are greater than one for women in all groups, and consistently close to
one for other applicants. Again, minority and women winners appear to generally secure
licenses with a greater share of the total value of licenses than their share of the upfront
payments.
The same pattern is repeated in the analysis of return-on-payment ratios by company size
(Table 24): minorities and women have return-on-payment ratios in excess of one, while
for other applicants the ratios are close to one.
The analysis of the return-on-payment ratios illustrates that among qualified applicants,
minorities and women win licenses of generally the same, or higher, value than other
applicants, relative to their upfront payments.
Table 1
General Utilization Ratios of All Applicants by Auction
+ Not enough minority or women winners or applicants for valid chi-square test.
* Significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Auction group 1: With installment plans Auction group 2: Without installment plans
Industry group 1: Advanced paging/data Industry group 2: Mobile voice&data Industry group 3: Interactive data Industry group 4: Wireless cable
Industry group 5: Multichannel video
Table 2
General Utilization Ratios of All Applicants by Auction Group
- Auction group 1 = With installment plans (consists of auctions 2, 3, 5, 6, 7, 10, 11).
- Auction group 2 = No installment plans (consists of auctions 1, 4, 8, 9, 12, 14, 15, 16, 17, 18, 20, 21).
* Statistically significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Table 3
General Utilization Ratios of All Applicants by Industry Group
- Industry group 1 = Advanced paging/data (consists of auctions 1, 3, 18).
- Industry group 2 = Mobile voice and data (consists of auctions 4, 5, 7, 10, 11, 12, 14, 15, 16, 20 , 21).
- Industry group 3 = Interactive data (consists of auctions 6, 17).
- Industry group 4 = Wireless cable (consists of auctions 8, 9).
- Industry group 5 = Multichannel video (consists of auction 2).
Table 4
General Utilization Ratios of All Applicants by Company Size
+ Not enough minority or women winners or applicants for valid chi-square test.
* Significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Table 5
Qualifying Ratios By Auction
+ Not enough minority or women winners or applicants for valid chi-square test.
* Statistically significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Auction group 1: With installment plans Auction group 2: Without installment plans
Industry group 1: Advanced paging/data Industry group 2: Mobile voice&data Industry group 3: Interactive data Industry group 4: Wireless cable
Industry group 5: Multichannel video
Table 6
Qualifying Ratios by Auction Group
- Auction group 1 = With installment plans (consists of auctions 2, 3, 5, 6, 7, 10, 11).
- Auction group 2 = No installment plans (consists of auctions 1, 4, 8, 9, 12, 14, 15, 16, 17, 18, 20, 21).
* Statistically significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Table 7
Qualifying Ratios by Industry Group
- Industry group 1 = Advanced paging/data (consists of auctions 1, 3, 18).
- Industry group 2 = Mobile voice and data (consists of auctions 4, 5, 7, 10, 11, 12, 14, 15, 16, 20 , 21).
- Industry group 3 = Interactive data (consists of auctions 6, 17).
- Industry group 4 = Wireless cable (consists of auctions 8, 9).
- Industry group 5 = Multichannel video (consists of auction 2).
Table 8
Qualifying Ratios by Company Size
+ Not enough minority or women winners or applicants for valid chi-square test.
* Statistically significant at the 95% level.
^ See Appendix A for details on chi-square tests.
** Auctions where women or minorities did not participate are excluded from weight calculations.
Table 9
Success Ratios of Qualified Applicants by Auction
+ Not enough minority or women winners or applicants for valid chi-square test.
* Statistically significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Auction group 1: With installment plans Auction group 2: Without installment plans
Industry group 1: Advanced paging/data Industry group 2: Mobile voice&data Industry group 3: Interactive data Industry group 4: Wireless cable
Industry group 5: Multichannel video
Table 10
Success Ratios of Qualified Applicants by Auction Group
- Auction group 1 = With installment plans (consists of auctions 2, 3, 5, 6, 7, 10, 11).
- Auction group 2 = No installment plans (consists of auctions 1, 4, 8, 9, 12, 14, 15, 16, 17, 18, 20, 21).
* Statistically significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Table 11
Success Ratios of Qualified Applicants by Industry Group
- Industry group 1 = Advanced paging/data (consists of auctions 1, 3, 18).
- Industry group 2 = Mobile voice and data (consists of auctions 4, 5, 7, 10, 11, 12, 14, 15, 16, 20 , 21).
- Industry group 3 = Interactive data (consists of auctions 6, 17).
- Industry group 4 = Wireless cable (consists of auctions 8, 9).
- Industry group 5 = Multichannel video (consists of auction 2).
Table 12
Success Ratios of Qualified Applicants by Company Size
* Statistically significant at the 95% level.
** Auctions where women or minorities did not participate are excluded from weight calculations.
^ See Appendix A for details on chi-square tests.
Table 13
Economic Value Ratio by Auction
Auction group 1: With installment plans Auction group 2: Without installment plans
Industry group 1: Advanced paging/data Industry group 2: Mobile voice&data Industry group 3: Interactive data Industry group 4: Wireless cable
Industry group 5: Multichannel video
Table 14
Economic Value Ratio by Auction Group
- Auction group 1 = With installment plans (consists of auctions 2, 3, 5, 6, 7, 10, 11).
- Auction group 2 = No installment plans (consists of auctions 1, 4, 8, 9, 12, 14, 15, 16, 17, 18, 20, 21).
Table 15
Economic Value Ratio by Industry Group
- Industry group 1 = Advanced paging/data (consists of auctions 1, 3, 18).
- Industry group 2 = Mobile voice and data (consists of auctions 4, 5, 7, 10, 11, 12, 14, 15, 16, 20 , 21).
- Industry group 3 = Interactive data (consists of auctions 6, 17).
- Industry group 4 = Wireless cable (consists of auctions 8, 9).
- Industry group 5 = Multichannel video (consists of auction 2).
Table 16
Economic Value Ratio by Company Size
Table 17
Average Economic Value per Winner by Auction
* Statistically significant at 95% level.
+ Not enough minority or women winners or applicants for valid t-test.
^ See Appendix B for details on t-tests.
Auction group 1: With installment plans Auction group 2: Without installment plans
Industry group 1: Advanced paging/data Industry group 2: Mobile voice&data Industry group 3: Interactive data Industry group 4: Wireless cable
Industry group 5: Multichannel video
Table 18
Average Economic Value per Winner by Auction Group
- Auction group 1 = With installment plans (consists of auctions 2, 3, 5, 6, 7, 10, 11).
- Auction group 2 = No installment plans (consists of auctions 1, 4, 8, 9, 12, 14, 15, 16, 17, 18, 20, 21).
* Statistically significant at the 95% level.
^ See Appendix B for details on t-tests.
Table 19
Average Economic Value per Winner by Industry Group
- Industry group 1 = Advanced paging/data (consists of auctions 1, 3, 18).
- Industry group 2 = Mobile voice and data (consists of auctions 4, 5, 7, 10, 11, 12, 14, 15, 16, 20 , 21).
- Industry group 3 = Interactive data (consists of auctions 6, 17).
- Industry group 4 = Wireless cable (consists of auctions 8, 9).
- Industry group 5 = Multichannel video (consists of auction 2).
Table 20
Average Economic Value per Winner by Company Size
*Statistically significant at the 95% level.
+ Not enough minority or women winners or applicants for valid t-test.
^ See Appendix B for details on t-tests.
Table 21
Return on Payment Ratios for Qualified Applicants by Auction
No data indicates that no data were available for these categories.
Not included indicates that data exist for these categories but have been omitted so that totals are calculated for auctions with valid upfront payment data only.
Auction group 1: With installment plans Auction group 2: Without installment plans
Industry group 1: Advanced paging/data Industry group 2: Mobile voice&data Industry group 3: Interactive data Industry group 4: Wireless cable
Industry group 5: Multichannel video
Table 22
Return on Payment Ratios for Qualified Applicants by Auction Group
- Auction group 1 = With installment plans (consists of auctions 2, 3, 5, 6, 7, 10, 11).
- Auction group 2 = No installment plans (consists of auctions 1, 4, 8, 9, 12, 14, 15, 16, 17, 18, 20, 21).
Table 23
Return on Payment Ratios for Qualified Applicants by Industry Group
- Industry group 1 = Advanced paging/data (consists of auctions 1, 3, 18).
- Industry group 2 = Mobile voice and data (consists of auctions 4, 5, 7, 10, 11, 12, 14, 15, 16, 20 , 21).
- Industry group 3 = Interactive data (consists of auctions 6, 17).
- Industry group 4 = Wireless cable (consists of auctions 8, 9).
- Industry group 5 = Multichannel video (consists of auction 2).
Table 24
Return on Payment Ratios for Qualified Applicants by Company Size
No data indicates that no data were available for these categories.
Not included indicates that data exist for these categories but have been omitted so that totals are calculated for auctions with valid upfront payment data only.
Appendix A: Chi-Square Test of Statistical Significance
A chi-square test is used to determine whether there are systematic differences in
utilization ratios across demographic groups for each auction. The null hypothesis of the
chi-square test is that there are no such differences in ratios.
The chi-square test statistic is based on a frequency table. In the case of utilization ratios,
the frequency tables are 2x2 tables with demographic group (e.g. minority/non-minority)
along one dimension and auction outcome (e.g., winner/non-winner) along the other. The
chi square test statistic is calculated as:
*df2 =
where fij is the observed frequency in the ij cell of the frequency table
eij is the expected frequency in the ij cell if the null hypothesis is true
R is the number of rows in the frequency table
C is the number of columns in the frequency table
The expected frequency under the null hypothesis is calculated as:
eij =
where fi is the total in the ith row marginal
fj is the total in the jth row marginal
N is the total number of observations in the frequency table
The degrees of freedom for this test statistic are given by:
df = (R-1)(C-1)
Reference:
Knoke, David, and George W. Bohrnstedt, Statistics for Social Data Analysis, F.E.
Peacock Publishers Inc., Itasca IL.,1994.
Appendix B: t test: Testing for the Difference Between Two Means
A t-test is used to determine whether two means are equal. The null hypothesis under the
t-test is that there are no differences in the means. To determine if the difference of the
average net revenues between the minority and non-minority license winners and the
male and female winners were significantly different, we used the following form of the
t-test:
tdf =
where is the mean of the first group
is the mean of the second group
S12 is the variance from the first group
S22 is the variance of the second group
n1 is the total number of observations in the first group
n2 is the total number of observations in the second group
The degrees of freedom can be approximated as follows:
df =
The null hypothesis that the mean difference is zero is rejected if the test statistic t is
greater than the critical value at a .05 level of significance. In other words, if the
calculated t is greater than the critical value, we would conclude that there is a statistically
significant difference between the two means.
Reference:
Dixon, Wilfrid J., and Frank J. Massey Jr., Introduction to Statistical Analysis, McGraw-
Hill Company, New York, 1969.
Throughout this report, an auction applicant is defined as a company (or an individual) who submitted
Form 175 (“Short Form”) indicating an interest to participate in an auction. In order to bid in an auction,
applicants had to qualify by paying an upfront payment. Not all applicants qualified to bid.
As measured by the FCC’s net revenue. Net revenue is the revenue collected from the winning bidder,
after deducting bidding credits, if any.
For the purposes of this report, “economic value of a license” is defined as the net revenue the FCC
received for the license. Net revenue equals the winning bid, minus bidding credits, if any.
See pages 9-10 for a discussion of the concept of statistical significance.
Calculated across all auctions.
Advanced paging/data; mobile voice and data; interactive data; wireless cable; multichannel video.
Auctions 1-21, with the exception of auctions 13 and 19, which were never held.
E&Y has not conducted an independent audit of the data and makes no representations of the accuracy of
the data.
The type of verification has varied over time.
In the very first auctions (Auctions 1 through 3), minority and women owned businesses were granted
bidding credits, but subsequently, only small businesses have qualified for bidding credits. Hence, in most
auctions applicants have not had a direct reason to identify themselves as minority- or women-owned
businesses. On the other hand, bidding credits for small businesses have been in place in many auctions
(Auctions 3, 4, 5, 6, 7, 10, 11, 14, 16, 17, 18, 20, 21).
The auction rule structures groups are based on the information provided to us by the FCC. The industry
groups are based on our understanding of the information in the bidder packages provided to us by the FCC.
For the purposes of this report, a company is considered large, if it has not identified itself as a small
company on the auction application.
It should be noted, though, that the category ‘large’ may encompass very wide differences in company
size, given the way it has been defined.
Auctions or auction groups without any minority/women applicants are omitted from these averages.
A Chi-square test was applied when the ratios were based on counts, such as number of auction winners.
A t-test was applied when the ratios were based on continuous variables, such as revenue amounts.
Appendices A and B describe the calculation of these statistics in detail.
Recall that for the purposes of this report, economic value of a license is defined as the winning bid for
that license, minus any bidding credits, i.e., the FCC’s net revenue for the license.
No statistical tests of significance are reported for these shares. If the minority share of economic value is
X% then the non-minority share is by definition (100-X)%, and testing for statistical significance in the
shares is not meaningful.
Other factors, such as ability to raise capital for upfront payments, may also play a role in the amount of
the upfront payment.
17
17
12/05/00
37
10/05/00