30/04/2026
The backward-looking bias in spectrum price benchmarking
Many regulators and operators use spectrum price benchmarking to estimate current market value. This can be a useful tool for regulators when setting reserve or renewal prices and to provide context for operators carrying out a spectrum valuation exercise.
However, if conducted without care, benchmarking can be misleading. Correct interpretation of benchmarks requires an understanding of spectrum price trends – without it comes the risk of significantly overestimating spectrum value.
Benchmarking is inherently backward-looking
Fundamentally, spectrum price benchmarks tell you what the price of spectrum was at a certain point in time. They are inherently backward-looking.
To make historical benchmarks useful, adjustments are often made for example to account for population, GDP and licence duration. However, another key question is ‘what does this price mean today?’.
When setting spectrum prices – for example, reserve prices for auctions – regulators often adjust historic benchmarks upwards to reflect the inflation that has occurred since the award. They then take a median of the benchmarks – implicitly assuming that spectrum prices have remained flat in real terms. But is this a reasonable assumption?
Spectrum prices have clearly fallen
Figure 1 below shows a three-year moving average of spectrum prices (in real terms) from spectrum awards globally across four band groups. It is taken from a publicly available dataset compiled by the Australian regulator ACMA1:
Two clear trends can be observed:
- Before 2015, spectrum prices broadly increased
- Since 2015, prices have decreased, particularly in sub-1GHz and lower 1-3GHz (1800MHz & 2.1GHz) bands.
To analyse this more formally, statistical tests – such as the Mann-Kendall test – can be applied to determine whether there is significant evidence of a decrease in prices over 2015-2025. Using this test gives strong evidence of a decreasing trend in sub-1GHz and lower 1-3GHz prices. Therefore, it is not reasonable to assume that spectrum prices have remained flat.
Furthermore, this downward trend is widely recognised in the industry. Leading regulators such as ACMA (Australia) and Ofcom (UK) have acknowledged the general decline in spectrum prices in recent years:
- In its ‘Review of Annual Licence Fees’, Ofcom notes “the long-term trend of declining spectrum values”2.
- Similarly, in its recent consultation on expiring spectrum licences, ACMA observes “long term price trends, as earlier awards were materially higher than recent outcomes” for sub-1GHz and lower 1-3GHz spectrum3.
The problem of ignoring the trend
If the trend is ignored – how big is the problem? For an illustrative example, assuming that spectrum prices have decreased by ~15% per year (real terms) for the past ten years – which is reasonable in particular for lower frequency bands – could lead to ~120% overestimate of market value. The figure below shows graphically the scale of this potential overestimate.
How to measure and account for the trend
Given the potential for a very significant overestimate of market value – what should regulators and operators do to avoid this error?
Firstly, the trend should be quantified. For example, where there is statistically significant evidence of a trend (e.g. using the Mann-Kendall test), the trend can be quantified using e.g. the Sen’s slope estimator (preferred) or standard linear regression. This shows the annual percentage change in prices each year (by measuring the linear decrease in log prices).
Having quantified the trend, historical benchmarks can be adjusted accordingly – in a similar fashion to inflation adjustments. Taking a median of the adjusted values yields a better estimate of market value that accounts for the spectrum price trends observed.
Conclusion
Ignoring trends in spectrum prices risks a benchmarking exercise giving significant overestimates of market value. This can be detrimental when estimates are used by regulators to set renewal or reserve prices for spectrum awards.
Prices that are an overestimate of market value may result in spectrum going unsold, meaning this valuable public resource is underutilised. Alternatively, bidders may overpay for spectrum which, in the case of mobile operators, for example, hampers their ability to invest in networks improvements and leads to poor service quality.
To avoid such issues, it is crucial to assess whether there are trends in spectrum prices when using benchmarks. Measuring the trends and adjusting for them results in a more accurate estimate of market value.
[1] Ofcom, Review of Annual Licence Fees: Decision to revise Annual Licence Fees for 900, 1800 and 2100MHz spectrum, 18 July 2025, Paragraph 4.5.
[2] Australian Communications and Media Authority, Expiring spectrum licences, stage 4: Updated preliminary views on pricing, December 2025, p.29.
[3] www.acma.gov.au/consultations/2025-12/expiring-spectrum-licences-stage-4-updated-preliminary-views-pricing.
