Most Copenhagen Airbnb hosts did not report their income to tax authorities, resulting in large losses in tax revenue

Knowledge overview

January 2024

Externally reviewed

A new study from Copenhagen Business School, supported by the ROCKWOOL Foundation, estimates that more than 176 million DKK have not been reported to the Danish tax authorities by Copenhagen Airbnb hosts in 2018 – and that even an information sharing agreement between Airbnb and the Danish government in 2019 did not stem the tide.

Also in this knowledge overview:

Income under-reporting varies with hosts’ wealth, age, nationality, and gender

Download knowledge overview

The platform economy and the online, peer-to-peer transactions it has enabled are growing worldwide. Airbnb, currently present in more than 100,000 cities worldwide, is a notable example of this phenomenon. Platforms like Airbnb have allowed ordinary people to act “as businesses” by sharing underutilized resources with others for a fee.

However, without fitting regulation, tax authorities rely on the honest self-reporting of income by platform participants, which creates opportunities for under-re-porting the income they earn “behind the screen”. If so, a substantial amount of tax revenues could be lost. This study finds that Airbnb hosts who earned income above the tax-free threshold during 2018-2019 rarely declared any earnings, even when the platform started sharing information with the Danish tax authorities (SKAT) from July 2019.

Peer-to-peer activities pose new threats to tax compliance

Most individuals renting their home on Airbnb do so occasionally and do not earn enough to be subject to taxation. Indeed, using a sample of nearly 22,000 hosts active in the Copenhagen area in 2018 as an illustration (Figure 1), most of the income made on Airbnb rentals was tax-free – more precisely, 224 million out of the 321 million DKK estimated for that sample of hosts. Only 23% of those hosts were liable to pay taxes but 93% of them have not reported any rental income to SKAT.

As a result of this behavior, the authors estimate that in 2018 alone around 90 million DKK remained off the radar of SKAT within this sample of hosts. Given the larger scale of Airbnb activity in Denmark and the conservative approach used to estimate income from Airbnb rentals, the authors estimate much larger volumes of under-re-ported income by Airbnb hosts in all of Copenhagen (i.e. 176 million DKK in the baseline estimation scenario and as much as 508 million DKK in the least conservative scenario reported in the study). It is worth noting that the national volumes of income under-reporting were naturally higher because Copenha-gen represents only around 45% of the Airbnb activity in Denmark. Considering that the tax rate on capital income ranged from 27% to 42% in the same period, the study uncovers non-negligible losses in tax revenues prior to the regulation of Airbnb income reporting.

Note: Income earned on Airbnb ac-tivities in millions of DKK; estimations based on 21,675 hosts in the Copen-hagen area in 2018.

The effects of the reporting agreement between Airbnb and SKAT

From July 1st, 2019, Airbnb started sharing information with SKAT about hosts’ rental income, thereby increasing the risk of being caught when under-reporting. However, without automatic reporting in place (which was only introduced in Denmark in 2021), hosts still had the obliga-tion and thus leeway to report (or not) the rental income they made on the platform themselves.

After comparing the behavior of likely hosts in 2018 and 2019, the authors indeed found only modest differences. As Figure 2 illustrates, the volume of rental income un-reported by the average active host in the sample at the end of the year decreased by only 8% (from 6,022 DKK to 5,519 DKK). More than 87% of tax-liable hosts reported no income at all in 2019, but this share used to be even higher in the previous year (93%). At the same time, more hosts reported their income fully to SKAT, but these were still the minority – 3.9% of hosts in 2019, whereas only 2.7% in 2018.

After isolating the role of other factors that correlate with income (under)reporting, the authors still find sig-nificant differences in reporting behavior across the two years, but these were rather small: the share of taxable rental income that remained “hidden” (voluntarily or not) from SKAT only decreased by 2.6 percentage points (see Figure 3). Despite the expectation that such an interven-tion would prevent tax evasion, the impact of pure infor-mation sharing agreements seems minimal according to this study.

Note: “Av. Underreported Income” denotes the estimated rental income (left axis) that the average active host did not report to SKAT in the re-spective year. “No income reporting” represents the share (in %, right axis) of tax-liable hosts who did not report any rental income to SKAT in the re-spective year. “Full income reporting” denotes the share (in %, right axis) of tax-liable hosts who reported all the rental income earned on Airbnb.

Income under-reporting varies with hosts’ wealth, age, nationality, and gender

The study also shows that under-reporting behavior varied across hosts: for example, immigrants tend to be more tax-compliant than Danish hosts. Income under-reporting also varied with hosts’ wealth and demographics, being more prevalent among less wealthy, younger (under 36 years old) and, intriguingly, female hosts (see Figure 3).

Note: Marginal effects of selected individual attributes on the share of income earned on Airbnb but not reported to the tax authorities. Analyses based on the subset of hosts subject to taxation. Vertical bars represent the percentage points difference between groups (e.g., women versus men), controlling for other characteristics of hosts and their listings and clustering standard errors at the individual level. The effects of different age groups must be compared against hosts who were 25 or younger. Dark bars represent 95% confidence intervals. Wealth is measured in million DKK.

The authors detect other important dynamics in the platform after the introduction of the reform. There was an overall decrease in the number of hosts that were ac-tive on the platform in 2019, breaking a previous positive trend. Similarly, the number of hosts that earn enough to be tax liable decreased. This indicates a change in the population of active and tax-liable hosts, which has con-tributed to an overall increase in prices (see Figure 4a).

Importantly, these dynamics did not harm the average quality of Airbnb listings; instead, review scores increa-sed during 2019 (Figure 4b). Although the regulation did not have large immediate effects on income reporting, it has changed the supply side of the platform and apparently enhanced the match between guests and hosts, which might lead other platforms to enter similar agree-ments.

This study shows that the rise of the sharing economy, if not regulated, can intensify existing issues of tax evasion, even in countries with high tax morale and low corruption like Denmark. However, given the short timeframe avai-lable after the reform, the authors are cautious in deri-ving final conclusions regarding the effects of the Airbnb-SKAT reporting agreement. Since 2021, Airbnb income is reported to SKAT automatically. This calls for follow-up research on how subsequent regulations impacted the platform and its participants.

Note: Dark lines represent monthly coefficients, price changes are noted in percentages, quality changes are absolute differences in 5-star ratings. The underlying model controls for individuals’ demographics and listing characteristics. The orange horizontal lines indicate 6-month averages.

About the Study

Linking Airbnb Hosts to the Danish Population Registers

The study combines data for 2017-2019 from three major sources: scraped data on Airbnb listings from insideairbnb.com, the Danish building registry (BBR), and population registers maintained by Denmark Statistics (DST). The authors use the approximate geolocation of Airbnb listings provided by insideairbnb.com to identify all the potential addresses in the BBR registry, for which the occupants’ first name is then compared to the name of the respective Airbnb host. This process allows the authors to identify a single individual for 32% of the listings scraped – these constitute the “core sample” of the study and include 12,774 hosts. For the remaining 68% listings, the authors identify multiple individuals with the same first name as the host of the listing. To identify the individual that is most likely to be the true host, the authors use a binary machine learning classifier (XGBoost) trained on the core sample and all other potential candidates (who do not match by first name). This allows the authors to identify 9,269 additional hosts. The study reports results obtained with both the core and extended samples of hosts.

Key Analyses: Participation, Income Under-reporting, and Platform Metrics

The authors conduct three main sets of analyses. First, they study participation and estimate, for each individual living in Copenhagen, their probability of being a host on the Airbnb platform. Details about these analyses and the exact variables included in the estimation are described in the study.

Second, the authors leverage details about listing prices and reviews received to estimate the income earned on Airbnb by each host, as well as their tax liability (by subtracting the deductibles applicable in each fiscal year). Different income scenarios (described in the paper) are used to test the robustness of the results. Hosts’ tax liability was then compared to the rental income reported in their individual tax declaration. Income under-reporting is defined as the difference between the estimated taxable income and the rental income reported. The study estimates models that explain the volume and share of income under-reporting based on a rich set of host and listing characteristics. Beyond these, the authors estimate the change in income reporting from 2018 to 2019, to gauge the immediate effects of the information sharing agreement between Airbnb and SKAT.

Finally, the study includes several supplemental analyses testing the effects of the reform on platform turnover (hosts’ entry and exit) and a number of relevant metrics capturing platform activity, listing prices, and quality.