3. Digital imports: methods and results

3.1 Dwelling services due to digital intermediation

Import of dwelling services due to digital intermediation is currently a part of travel measured in the international trade in services statistics. To compile an estimate of the import of dwelling services through digital intermediation we created a model with several assumptions. We assume that:

• Dutch residents only rent from foreigners when they stay abroad. 
• Business travellers do not stay in Airbnb’s.
• In the period 2010-2016, Airbnb was the market leader in dwelling services due to digital intermediation. We assume only calculating the imports due to Airbnb is sufficient for estimating the dwelling services imported due to digital intermediation. 

Our research has led to the following additional information:

• We found partial market shares of Airbnb in the total lodging for Amsterdam (Colliers, 2017), US (Molla, 2019) and Europe (Kookie, 2019) in different sources. We found more sources than shown below, but most were in line with these numbers. There are some differences between the market shares of Europe and the US, but that seems logical as Airbnb started in the US. We compiled the below market share of Airbnb in the lodging of Dutch tourists abroad based on the market share of Europe as that is the destination for 80 percent of Dutch travel. The 2016 market share is estimated by taking both the doubling of the number of arrivals (Leonardo, 2019) as well as the doubling of the Amsterdam market share as leading. 

• To determine the part of the payment which is reserved for the dwelling service and the part paid as a fee for the intermediation, we need to know what percentage of a booking is paid to Airbnb. There are some different terms and conditions possible, but in general 18 percent (Airbnb, 2020) of the total paid amount is for the intermediation by Airbnb. 
• We use information from the Continuous Holiday Survey conducted among Dutch households to make estimates of the amounts they spend on travel. From 2015 onwards, this survey contains information on the amount paid for the accommodation and the type of accommodation travellers stayed in. For previous years we only have the total amount spent on the holiday. For 2015 and 2016, we calculated the amount spent on accommodations (hotels, B&Bs, privately owned houses, apartments and motels) per quarter set off against the total amount spent on the holiday. The ratios proved very stable between both years at 33 percent, 24 percent, 22 percent and 28 percent of the total. These ratios were then used on the totals spent on the holiday per quarter for the previous years. This results in a slowly increasing time series of total Dutch spending on accommodations abroad of around 3 billion euros.

Using this additional information we created an estimate for the dwelling services due to digital intermediation of Airbnb which consists of the revenue of foreign Airbnb vendors from Dutch travellers. This revenue started in 2011 with 2 million euros and quickly rose to 341 million euros in 2016. 

3.2 Digital intermediation services of hotel services

The import of digital intermediation services of hotel services is currently a part of travel measured in the international trade in services statistics. To compile an estimate of the import of hotel services via digital intermediation we created a model with several assumptions. 

We assume that:

• Business travellers do not book their stays via online platforms based outside the Netherlands. 
• The market leaders for digital intermediation of hotel services in Europe are Booking.com and Expedia with close to 90 percent of the market (HOTREC, 2020). We assume that other hotel services platforms use similar conditions for hotels and consumers to use their platform as these two do.

Our research has led to the following additional information:

• We use information from the Continuous Holiday Survey conducted under Dutch households to make estimates on the amounts they spend on travel. From 2015 onwards, this survey contains information on the amount paid for the accommodation and the type of accommodation travellers stayed in. For previous years, we only have the total amount spent on the holiday. For 2015 and 2016, we calculated the amount spent on hotel services per quarter set off against the total amount spent on the holiday. This proved very stable ratios between both years at 18 percent, 14 percent, 10 percent and 17 percent of the total. These ratios were then used on the totals spent on the holiday per quarter for the previous years. This results in a slowly increasing time series of total Dutch spending on accommodations abroad of a little less than 2 billion euros.
• To determine the part of the payment which is reserved for the hotel service and the part paid as a fee for the intermediation, we need to know which percentage of a booking is paid to the platform. There are some different terms and conditions possible, but in general 15 percent (Skift, 2020) of the total paid amount is for the intermediation by Expedia or Booking.com. As they are the dominant players in the market, we use this ratio for all platforms. 
• To determine the market share of platforms in the total of hotel services imported by Dutch travellers we used data collected by HOTREC (2018). HOTREC is the umbrella association of Hotels, Restaurants, Cafés & similar establishments in Europe. They have data for 2013, 2015, 2017 and 2019 on the market share of platforms in the booking of hotel services in Europe. For the years for which we have data, we see a steady increase of the market share. We therefore assume that the missing years hold the middle between the values for 2013 and 2015 for 2014, and respectively the middle between the values for 2015 and 2017 for 2016. For the years before 2013, no reference was found. We did look into the development of the turnover for Booking.com, but this did not seem completely representative for the Netherlands, as Booking.com started here. We ended up using the 2013/2014 growth as a proxy for the year-on-year growth from 2010-2013.
• The HORTEC data also has a division to exact platforms in Europe for the booking of hotel services. The latter is necessary as Booking.com, the market leader with the share of 63.5 percent in 2015, is a Dutch-based company and therefore the intermediation service is not imported. We used the market share for Expedia (17.7 percent in 2015), HRS (12.8 percent in 2015) and their market share in the total on the non-specified platforms (7.3 percent, of which we assume 32 percent is imports). 

Based on this additional information we created an estimate for the import of digital intermediation services of hotel services which consists of the revenue earned by foreign platforms by supply of intermediation of hotel services to the Dutch travellers. We estimate that this revenue at 17 million euros in 2010 has risen to 22 million euros in 2017. 

3.3 Upscaling taxi services due to digital intermediation

Import of taxi services due to digital intermediation is currently a part of travel measured in the international trade in service statistics. However, it is only a part in a larger category which proved impossible to split. To compile an estimate of the import taxi services due to digital intermediation we created a model without using ITSS. This model holds several assumptions. 

We assume that:

• Uber only became relevant in Europe in 2013 (CBSinsights, 2020). For the preceding years, we assume that no taxi services via intermediated platforms were imported by Dutch tourists abroad. 
• For the 2010-2016 period, Uber was by far the dominant market player. We assume their data is therefore applicable for all taxi services due to digital intermediation. 
• People who used an Uber on their holiday in the earlier years (2012-2014) when this was not commonplace are “early adapters” who are likely to do the same on a next holiday. On average, people who say they have ordered a taxi via digital intermediation, will most likely have done so more than once in a year as people often need to get to and from a certain place such as the airport and people can have multiple holidays in one year. We make an expert guess that people who have ordered a taxi via digital intermediation in a year will have taken 2 Uber rides. 
• For this estimate we use data from the Continuous Holiday Survey conducted among Dutch households. The part of this data we use here was collected by asking people who had travelled for business purposes in 2016 on their use of digital intermediation to order taxi services on personal holidays. We assume that the estimate for this subsample of the Dutch residents is applicable to all Dutch residents who go on holiday. 
• People who go on business trips will also go on a personal holiday in that year. 
• People generally go on holiday with their household.
• We leave business-related travel out of this research as we assume that Uber use during a business-related stay abroad will be paid by the enterprise and thereby it will be included in the business statistics and not as a personal expenditure.

Our research has led to the following additional information:

• We use information from the Continuous Holiday Survey conducted among Dutch households to make estimates on the amounts they spend on business-related travel. For 2016, extra questions were added on the use of platforms. One of the questions to respondents was whether they had ordered a taxi via digital intermediation on personal holidays abroad. The results of this survey in 2016 were that 7 percent of the grossed up sample indicated to have ever used digital intermediation to order a taxi on a personal holiday abroad (CBS, 2017b). 
• To determine the annual growth of digital intermediation on taxi services we use the revenue of Uber. This has grown from 0.1 billion in 2013 to 6.5 billion in 2016 (Statista, 2020). 
• As we know the share of people who have used digital intermediation to order a taxi, we need to know how many households there are in the Netherlands. This we know from our own data (CBS, 2020c).
• We also need the CBS information on the share of the population that goes on holiday, as we assume that business travellers go on personal holiday and that their use of digital intermediation on taxi services is representative for other Dutch travellers abroad on personal holidays. Roughly 80 percent of the Dutch population go on holiday each year (CLO, 2017). 
• Finally we need to know how much a Dutch tourist pays a taxi driver which they booked through digital intermediation. We use information on what an Uber driver makes on average from a ride. We found this to be 15 euros. 

Combining this information we come to the following model for the share of Dutch citizens who go on holiday and who use digital intermediation to book a taxi abroad. We use the distribution of the revenue of Uber in the 2013-2016 period to distribute the 7 percent who ever used digital intermediation to book a taxi according to CBS (2017b). Next we use the assumption for the period 2013-2014, that people who book taxis via digital intermediation are loyal customers and that they will use this service in each flowing year as well. This results in an estimate of 0.08 percent in 2013 of the Dutch population who goes on holiday and has booked a taxi through digital intermediation. This share of the Dutch population has grown to 5.7 percent in 2016. 

3.4 Digital intermediation of travel (not taxi) services

Import of digital intermediation services of travel (not taxi) services is currently a part of travel measured in the international trade in services statistics. Digital intermediation of travel (not taxi) services focuses on the parts in travel which lend themselves for digital intermediation, which we have not yet studied in this grant. This consists of:

• Foreign digital intermediation on overnight stays in bungalow parks or campsites, boats or youth hostels (we will use “other accommodations” from here onwards). 
• Foreign digital intermediation on car rental.
• Foreign digital intermediation on excursions and tickets.

In order to make estimates on this digital intermediation we use several sources:

• We use information from the Continuous Holiday Survey (CHS) conducted under Dutch households to make estimates on the amounts they spend on travel. From 2015 onwards, this survey contains information on the amount paid for the accommodation and the type of accommodation travellers stayed in, local transportation (including rental cars, petrol and public transportation) and how much they spend on excursions and tickets. For 2010-2014, we only have the total amount spent on the holiday. For 2015 and 2016, we calculated the amount spent on these services per quarter set off against the total amount spend on the holiday. These ratios proved rather stable. The ratios were then used on the totals spent on the holiday per quarter for the previous years. This results in a slowly increasing time series of total Dutch spending on overnight stays in other accommodations, local transportation and tickets and excursions.

• Only air travellers rent cars abroad
• For 2012, we use the 35 percent as mentioned in the research. This also stated that in 2011 this was 2 percent less. We use this growth to cast back the ratio for 2010 and 2012-2016 share of air travellers that rent a car. 
• In 2018, the Trend report of Sunny Cars tells us that 58 percent of the costumers rent the car for the entire holiday, 35 percent for more than one day, on average half of the days and 7 percent rent a car for one day. In general this research indicates that people who rent a car rent it for close to three-quarters of the holiday. We use this for our estimates as Sunny Cars is the market leader in the Netherlands.  
• There are a lot of different prices for rental cars, but we assume a rental car costs about 15 euros per day.

We combine these estimations with information of the Continuous Holiday Survey on the number of people who went on holiday by air and the number of days they stayed on holiday. This results in the following model where imports of car rental are about half of the total local transport costs we saw in table 3.4.1, which seems plausible. 

3.5 Database services and database originals imports

The import of database services and database originals by enterprises is currently a part of the international trade in services statistics. Imports of database services and database originals by consumers is not yet captured in regular statistics.

Results

The estimation of imports of database services and database originals is presented in table 3.5.2. The value of the imported database services almost doubles in the 2010-2016 period, from 1.59 billion in 2010 to close to 3 billion in 2016. We have to make some remarks with respect to presented in table 3.5.2 figures. We deliberately did not use the category “Computer services, excluding ownership rights” for estimation database services, as it exceeds the definitions of them. Unfortunately we cannot estimate the probable share of database services in the category “Computer services, excluding ownership rights” by manually extraction or by any individual research due to complicated subject matter and a number of enterprises reporting “Computer services, excluding ownership rights”. On the other hand, we are aware that many companies will report cloud services under software licences as nowadays cloud services are often provided via “Software-as-a service” agreement. Taken all this into account we present below the estimated value of database-related services and database originals set off against the total value of the so called ICT-enabled services: “Computer services, excluding ownership rights”, “Computer software licences to use”, “Computer software licences to reproduce and distribute” and “Other information services” in our ITSS survey.

Enterprises

Following EBOPS 2010 - CPC 2.0 detailed correspondence table [1], licences to reproduce and/or distribute computer software comprise: Licensing services for the right to use computer software and databases as well licensing services for the right to reproduce, distribute or incorporate databases (i.e. compilations of facts/information) in other databases or applications. This applies to various levels of licensing rights:

• rights to reproduce and distribute the database
• rights to use database components for the creation of and inclusion in other databases and applications.

In the Dutch ITSS business survey we ask respondents to report the following categories which corresponds with the EBOPS definitions:

1. Licences to reproduce and/or distribute computer software (SH3)
2. Licences to use computer software (SI21Y)
3. Other information services (which includes database services and database originals) (SI32).

Making the assumption that the database services and related transfer of ownership rights through licensing are intrinsically included into above-mentioned questions, we will calculate the B2B part of import of database services (including database originals) based on above mentioned questions (1), (2) and (3) from our ITSS survey. We first selected all 744 enterprises that reported to have imported either (1), (2) or (3) in the years 2014-2016. Per type of service (SH3,SI21Y,SI32), we determined which enterprises contributed the most to the total of the service and we manually checked the largest enterprises to determine if their type of services could be classified as the import of database services or database originals. We checked a minimum of 67 percent of the value per service. The results of the checked enterprises can be seen in table 3.5.1 below. We used these ratios on the total value of the service per year to determine the value of the import of database services and database originals. 

Consumers

We tried to make an estimate of Dutch import of database services and database originals by consumers by using data from several external sources on Dutch consumers. This unfortunately did not result in plausible outcomes. To come to an acceptable estimate we modelled German estimations (Meinusch & Hessel, 2019) made by the Deutsche Bank in 2019 on digital trade of German households. In this research Deutsche Bank measures digital economy where their estimates specified by several different services including cloud services. We believe these are the types of services we are looking for to estimate Dutch consumers imports of database services and database originals bought outside the Netherlands. We make the German estimates suitable for Dutch consumers by correcting their estimate for the difference in population (Eurostat, 2020). 

Combined

In total we estimate that the imports of database services and database originals of Dutch enterprises and consumers adds up to approximately 3 billion euros in 2016. 

3.6 Imports of goods via e-commerce as a sales channel

The import of goods via e-commerce is part of the international trade in goods statistics if it is imported from outside the EU and has a value of 22 euros or more. Smaller amounts and imports of goods from other EU-countries are not registered in a statistic, but an estimate is made for the national accounts. We try to make a new estimate here. To compile an estimate of the import of goods via e-commerce we created a two-part model. First, we estimate the import of goods via e-commerce by consumers and second we estimate the same for enterprises. 

Consumers

Our research has led to the following additional information:

• Research done by Thuiswinkel.org (2017) gives us for the years 2010-2016 the amount that Dutch consumers spent on shopping online. For 2013-2016, this research also provides us with information on their spending on foreign websites. For 2016 and 2017, information is available on the share of online shopping that is spent on goods versus services. 

From this research we made the following model in which 2017 was necessary for the back cast of the share of goods in the total spending on foreign websites. The estimated shares, made by back casting, are in grey and italic. We estimate that 3.19 percent of total goods and services bought online in 2015 were bought from a foreign website. Of this share of 3.19 percent, we knew that 61 percent were goods. This resulted in an estimate of 313 million euros spent on goods imported via e-commerce (CBS, 2020d).

Enterprises

Our research has led to the following additional information:

• Statistics Netherland asked enterprises with 10 or more employees about their IT use (CBS, 2020e), this includes e-commerce. From 2013 onwards, questions are included on which part of their purchases are done via e-commerce (around 7 percent) and what part of those purchases are done abroad (around 10 percent). For earlier years, the questions are slightly different and do not include if the purchases are done abroad. 
• To split the purchases in goods and services we need to use the division between goods and services in the imports. For most years, this is little over 70 percent of the total imports (CBS, 2020f). 

To compile an estimate of the goods imported by enterprises through e-commerce we created a model with several assumptions. We assume that:

• Enterprises with fewer than 10 employees use e-commerce in similar ways in their imports as enterprises with 10 or more employees.
• The IT use questionnaire only asks enterprises if they purchase from e-commerce channels and if that purchase is within 8 ranges. To come to an average purchase from e-commerce channels we assumed that for all the enterprises who indicate a certain range we can use the mean of that range. For the last range, more than 50 percent of the purchases are done via e-commerce. We assume 60 percent for all those enterprises. This results in the non-coloured percentages of e-commerce in the total purchases. 

• In the table above, 2011 is very different from the other years. We therefore assume this is not the best estimate for 2011 and we replace this estimate with the 2011b one, which holds the middle between 2010 and 2012. 
• For 2013-2016, we have information on the part of e-commerce purchases that are done abroad. For 2010-2012, we assume that we can safely use the 2013 percentage as the 2013-2016 percentages are rather consistent.
• We need information on the total purchases of enterprises in the Netherlands. This information is not available, partly due to security reasons. We do have the total turnover of the business economy (NACE B-N, excl. K, incl. 95) for 2011-2016 period (CBS, 2020g), turnover for 90 percent of the business economy for 2010-2016 and the corresponding purchases (CBS, 2020h). We assume that the part of the purchases in the turnover for 90 percent of the business economy per year is a good proxy for the part of the purchases in the total business economy. 

• We also assume that the business economy is a good proxy for the total e-commerce purchases abroad. 

Combining this information we come to the following model. For 2016, close to 7 percent of the purchases was done via e-commerce of which almost 10 percent was done abroad and of those, 72.3 percent were goods. This totals to 0.4 percent of all purchases is a good bought through e-commerce abroad. Of the estimated 9 billion purchases in our business economy this results in 4.3 billion euros as our estimate of the goods purchased by enterprises via foreign e-commerce. This is an increase from 2010, but not a steep one.

Combined

In total we estimate that the purchases of Dutch enterprises and consumers of goods via e-commerce abroad adds up to close to 5 billion euros in 2016. 

[1] Correspondence between the EBOPS 2010 and the Central Product Classification (CPC, version 2) - Detailed version Consulted from: https://unstats.un.org/unsd/tradeserv/tfsits/msits2010/ebops2cpc_detailed.htm