Spectrum Pricing Explained

Author: Pau Castells, Head of Economic Analysis

Over the last decade, our use of streaming, scrolling, and video calls has exploded. To ensure that mobile networks can keep up, operators have invested in 4G and 5G infrastructure and acquired the rights to use additional radio frequencies. With more radio spectrum, users experience greater network quality and speeds.

However, market conditions have also changed in this period, but at a different rate than the prices that are charged for spectrum. As shown in our recent Global Spectrum Pricing report as well as in previous empirical research[1],  this poses problems for both consumers and operators. If operators need to spend more of their revenues on spectrum, that means less money for investment, which means poorer network coverage and slower speeds for everyone. 

Let us start from the beginning… 

How much more spectrum did operators get for 4G and 5G?

The amount of spectrum assigned to mobile networks has nearly doubled since 2014. This increasing trend has been seen across all regions, but there is a lot of variation in the total available spectrum. On average, a country in Latin America and the Caribbean or Sub-Saharan Africa, for example, has assigned about 400 MHz, while in Europe and in North America, many countries have already assigned close to 1,000 MHz.[2] This reflects the different stages in 5G rollout and adoption of mobile, as well as local regulator’s ability to make spectrum available.

Just as increasing data allowances for users has not resulted in an increase in overall consumer prices, increasing spectrum holdings does not automatically need to result in an increase in operators’ costs. However, that is exactly what has happened. Ongoing spectrum costs have increased as a percentage of revenues from 4% of operators’ recurring revenue in 2014 to about 7.4% in 2023. This is explained by several factors, starting with monetisation challenges.

Have mobile operators been able to monetise on the boom in mobile use?

Here’s where it gets interesting. One might think that operators rolling out 4G and 5G could capture some commercial value brought by the latest mobile use cases they support. And, indeed, the commercial value of the mobile and internet ecosystem increased 15% per year in the last decade and has reached almost $7 trillion, thanks to the rise of video calls, mobile banking, gaming and other applications. The value to consumers is likely to have been even higher. 

However, in actuality, very little of this value has been captured by mobile operators.  

Adjusted for inflation, average revenue per connection decreased by close to 40% over the past ten years. Each MHz of spectrum supports only a third of the revenue it supported a decade ago. For operators, who need more spectrum to deliver services customers pay less for, the commercial value of each chunk of spectrum has dropped. 

So, are spectrum unit prices reflecting this drop in commercial value?

To some extent, yes. But not sufficiently. 

We calculated what operators paid for each MHz and adjusted for the size of the market and inflation to allow for comparisons across markets and time. What we found is that unit prices (i.e. the price paid for each MHz) have been sliding, especially in the sub-1 GHz band: down almost 75% since 2012. These bands offer good coverage, but not enough capacity to bring 100+ Mbps to densely-populated cities. The ability to serve high-density areas is what higher bands can provide, but their prices have not fallen as much. Overall, spectrum prices are trending down, but not at the same rate as operator revenues. 

Wait, if prices are dropping, why is the total cost going up?

Even though each MHz is cheaper, operators are buying more spectrum to meet consumer needs. A good way to measure total cost is to sum it across all active licences and compare it to the revenues that can be obtained in the market with its use. But since spectrum investments are lumpy (i.e. they require a big chunk of cash at distant intervals), the cost of licences needs to be amortised over their duration. This way, we ensure that the upfront cost of a 10 or a 15-year licence is gradually expensed in even amounts over its duration. Once a licence expires or is renewed at no additional cost, it stops being expensed. This is a standard accounting procedure applied to other capital expenses, such as plant and machinery costs, which are amortised in the books over their lifespan. This gives a solid look at the ongoing expense. 

We find that ongoing spectrum costs have increased from 4% of operators’ recurring revenue in 2014 to about 7.4% in 2023. That’s a 63% increase over a span of a decade. To state the obvious, the overall cost increases if revenues decline at a faster rate than the per-unit price of spectrum falls and, as mentioned above, revenues have fallen. 

Importantly, we find large differences in this cost across countries. Yet, differences between operators within the same country tend to be much smaller. This points to policymakers being the driver - something quite obvious in countries where spectrum prices are set by regulators directly. However, in many countries pricing of spectrum is left to market forces and auctions so the influence of policy is not as obvious – we will come back to this later.

Should this matter to anyone but the MNOs and their investors?

High spectrum costs are certainly a problem for mobile operators, but have they impacted consumers, too? The results of our study show the answer is unambiguously ‘yes’. Governments that have set ambitious digital targets are also unlikely to see these being met where spectrum costs are high. 

Can anything be done about it?

Returning to policy, when prices are set directly by regulators, high spectrum prices are often a reflection of the regulator looking to recover some costs or raise revenue for the government.

However, regulators often rely on auctions as a method of putting spectrum to the best use, and the price is ultimately set by voluntary bids. This would seem like a smart, market-based decision. Yet, analysing thousands of licences, we find that decisions behind the scenes shape how much operators pay in auctions too. For example:

• Scarcity: While the laws of physics determine that the same amount of spectrum is available everywhere, it is regulators (as the single sellers of spectrum) who determine how much of it is made available to the market. If there’s not enough spectrum to go around, prices soar. In markets where the supply of spectrum is scarce, each unit of spectrum is more expensive.  
• Reserve Prices: High starting bids can leave spectrum unsold or lead to inflated costs. 37% of auctions ended at the reserve price, potentially exacerbating the ‘winners curse’ where the successful bid overvalues the asset – a well-established concept in economics.
• Coverage obligations: Increasingly regulators are imposing requirements for operators to deploy high-speed networks across large swathes of territory, even in remote areas that are not profitable for any operator. This is a noble goal, but clearly represents an additional cost which is often not reflected in the price attached to the license.

Undoubtedly, policy decisions wield considerable influence over spectrum costs. Even in instances where final prices are determined through auctions, foundational policies shape the outcomes significantly. 

Wrapping it up

Spectrum prices per MHz are declining, but total costs as a percentage of revenues have increased because operators need more of it AND revenues are falling faster than the price of spectrum. This is obviously a crucial matter for operators, but let us be clear: it is also a concern for regulators. Cost build-up can be detrimental to consumer outcomes such as coverage and speeds because it constrains operators’ ability to invest in networks. Fortunately, there is an opportunity to course-correct in the coming years. Between 2025 and 2030, close to 1,000 individual licences in 110 countries will expire, providing an opportunity to adjust spectrum prices to their actual market value. A rational approach to these renewals and new spectrum bands will help realise the full benefits of mobile technology.

For more insights, the full report, which is based on data from more than 250 operators in almost 100 countries, is available here.

[1]

 https://www.sciencedirect.com/science/article/pii/S0308596121001324

[2] This analysis excludes spectrum above 24 GHz (also known as mmWave spectrum)