Available data

The tool draws on multiple data sources, providing raw data for topics such as pricing, coverage, speeds, usage, adoption, industry financials, demographics and economic data. The tool uses these to create a set of output metrics. Some are pulled directly from the relevant underlying data while others involve minor recalibration. For example, underlying price figures are converted to affordability by dividing into GDP per capita. This ensures that for all metrics a higher figure is better. Data availability varies significantly. The range of metrics available for developed countries is much higher than that for developing countries.

Country sets

A country set is a group of counties, such as Africa, the OECD or high income. The spirit level tool allows the user to select which group comparisons should be applied to. This affects both the countries to be included and the benchmark for best performance within the country set. In practice, comparisons of a country to a relevant peer group are likely to be most helpful.

Metrics

Each spirit level is based on a set of three metrics, for example NGA coverage, affordability of 30-100 Mbps broadband and broadband competitive intensity. Each metric in a spirit level is rebased relative to the best and worst performers for that metric from the country set. The best performer sits at the outer end of the axis, and the worst performer at the inside end (the maximum and minimum values from the country set are provided in the key). For each metric, the median performance within the country set is shown by a dotted line.

Interpreting country plots

The area of a country’s triangle can be taken as a very crude measure of overall performance across the three selected metrics. Overall performance is clearly driven in part by GDP per capita, which enables a country to afford better outcomes. All else being equal, it is expected that dark red triangles (those for top quartile countries within the country set) would be larger, for example.

More importantly than the area of the plots, the skew can also be compared. A country may have much stronger performance in one metric compared to the other two – it as “skewed toward” that metric. Conversely, it may have much weaker performance in one compared to the other two – it is “skewed away” from that metric.

Spirit levels generally show great variation in skew. This suggests that the differences between countries are not simply a matter of wealth – if they were, one might expect the countries’ triangles to have different sizes but similar shapes.

Such skews may be by accident or design, and are not inherently good or bad. For instance, they may stem from a deliberate targeting of certain goals to gain an international competitive advantage. (Qatar’s FTTH national broadband network would be an example). They also may be driven by historical circumstances. For instance, Malta has long had good cable network coverage, which has supported a comprehensive NGA deployment.

Note that countries in the sprit level are ordered based on their skew, so that countries with similar skew are grouped together. A spirit level report can also suggest whether there is a trade-off between different metrics. For instance, if there are one or no countries that exceed the median on all three metrics, this suggests that the metrics may be in tension with each other.

Types of trade-offs

• Direct trade-offs are those between elements of communications, where there is a direct link between those elements. For example, a universal service obligation funded by an industry levy may improve coverage but conversely increase prices.
• Long-term trade-offs are those which may have immediate benefits, but which will likely carry costs in the longer term (or vice versa). For instance, a decision to impose stringent retail price controls will enhance affordability but might deter future investment.
• External trade-offs are those with consequences outside communications, and in particular those involving government funds. For instance, state subsidies for FTTH require a trade-off between the benefits of enhanced broadband vs other ways in which the government could spend its money.

Differences in trade-off decisions

• Different views of the future lead to different choices being made today. For example, some countries have anticipated an imminent need for ultrafast broadband speeds, and have subsidised FTTH/B. Other countries taking a different view of future bandwidth demand have not.
• Different industry starting points lead to different next steps. For example, if the incumbent has the only fixed network in a country, then local loop unbundling (LLU) may be essential for services competition. Conversely, if a country has widespread competition from cable networks then LLU may be unnecessary.
• Different economic and geographic starting points, such as widespread FTTH, might be simply unaffordable for a less developed country. Satellite broadband might be useful for a country with low population density, but not for a city-state.
• Different priorities will lead to different decisions. These priorities may relate to a specific issue (for example, Germany’s emphasis on data privacy) or broad philosophy (a preference for free market or statist approaches)

Existing research

Perhaps surprisingly, this is an area with relatively little existing research. Much literature focuses on only one element of the consequences of a given decision. For example, while it is useful to know that additional mobile operators keep prices lower, a regulator must consider the trade-off that those low prices and market fragmentation may deter investment. (See the full spirit level paper for a list of some academic research on regulatory decisions).

Mapping regulatory decisions and outcomes

Decision matrices have been created to explore trade-offs (one for fixed and one for mobile). For a range of decisions – such as “increase spectrum availability” or “apply NGA wholesale active access” – these matrices offer a view of the dimensions affected by that decision. For NGA active access, for example, the affected dimensions include NGA investment, competition, affordability, coverage and uptake.

For each of these dimensions, a tentative view is offered as to whether the impact is positive or negative. For instance, NGA active access might have a positive impact on NGA competition, but a negative impact on NGA investment. In some cases, the view is “+/-“, where the impact may vary across players or depend on local circumstances.

Note: For definitions of the decisions, see the spirit level paper

Using the matrices to identify “levers to pull”

The matrices have been built starting from decisions and considering the outputs affected. However, they can also be used in the opposite direction. By considering all the decisions that are linked to a given output, one can readily see the regulatory levers that are likely to be most relevant.

For example, if a regulator has a particular concern about the level of competition in its mobile market, the matrix highlights spectrum set-aside and availability and merger blocks as choices, which might improve matters on that particular dimension. (Though, as the matrix notes, these choices might have other adverse consequences).