Although roaming charges are strongly disliked by users, operators have long maintained that they cannot afford to remove them as they would lose profitable revenue and would have to raise domestic tariffs to compensate.
It is interesting to see that an increasing number of operators are now removing roaming charges. Last year, T-Mobile in the US announced that it was abolishing the charges altogether. Soon afterwards Three abolished roaming charges for users travelling between its different territories, and then for users visiting the US. This year Orange announced that it would also be abolishing roaming charges for calls between its territories in Europe during 2014, although initially only for some of its 4G users.
These announcements reminded me that a while ago I had carried out a price sensitivity analysis that showed that average revenue per user was typically higher for operators that offered lower average prices, since the level of increase in traffic was more than enough to compensate for any revenue loss. From these results I would expect operators that abolish roaming charges will enjoy increased revenues, especially given the number of users that avoid roaming altogether while travelling because of the cost.
I have set out a summary of this analysis here:
Price Sensitivity: Effect of Mobile Tariffs on User Demand and Revenues per User
Data on cost and usage of international roaming traffic, especially data traffic, is not widely available, but that for domestic traffic can be obtained more readily. Prices and usage for domestic mobile calls vary considerably between countries, which permit comparison of any relationships between the two.
Using data on voice ARPU and average called minutes from a large number of mobile operators worldwide, I prepared the chart shown here.
For each mobile operator it shows the average minutes of use per subscriber and the average price per minute. The trend line drawn through it shows that as the price decreases from €0.18 to €0.028, revenue increases from €10.80 to €22.40 as shown in the table.
International roaming charges are not shown, but in many cases will be well off the top of the left hand axis on the chart.
However, as many of the data points are bunched on the left of the graph, I decided to see if the data could be divided to help take account of the very different market conditions faced by operators. I found it helpful to split them into two groups based on their GDP/capita; developing and developed countries. Trend lines drawn through each group show a clear relationship between price and usage in line with typical price sensitivity curves.
Based on these curves, I produced two tables of price versus revenue, one for each group. For developed countries, the curve (in blue) shows a strong relationship between decreases in price and increased usage and average revenues per user. The figures are shown in the table below.
However the curve for developing countries (in red) shows a decrease in average revenues per user where prices are lowest and usage highest. This can be attributed to the much higher ratio between the higher average prices on the left of the chart and the lower ones to the right, and indicates that operators could go too far in reducing prices. In this case the ratio between the highest and lowest prices in the table is a factor of 26, compared to a factor of 4.7 for developed countries. Again the figures are shown in the table.
Altogether, the curves suggest that for domestic calls at least, optimum revenues are most likely to be achieved with prices in the range between around 4 and 7 Euro cents.
Of course this picture does not consider a number of more detailed factors, such as the additional network capacity that might be required to handle the traffic, nor the fact that in many developed countries handsets are subsidised by increased tariffs. Nevertheless, the principle can still be expected to apply. It would seem reasonable to assume that similar results would be produced by carrying out a similar analysis for international calls, preferably for both voice and data traffic, assuming the data were available. Given that many users avoid making calls while roaming in order to minimise the risk of bill shock, in practice the increase in usage could be markedly more dramatic.