Image credit: Unsplash

Strict Laws in Airbnb Singapore — An Analysis

Visualizing the proportion of listings that are challenging the law of short-term accommodation.

Singapore’s Law on Short-Term Stays

Apart from the lack of picture-worthy natural sites to explore and being one of the most expensive places to live, Singapore is undoubtedly one of the best cities for short-term travelling in Southeast Asia with the most efficient public transport system, affordable dining and clean environment.

Staying in an Airbnb property has always been the go-to option for many travelers who wish to explore places on a budget with a homely experience. However, there is a law enacted against the landlords in Singapore that imposes a fine up to $200,000 if guests stay for less than 3 months in a private property. Tougher laws apply to home-owners of HDB (Housing Development Board) estates, a more affordable property with limited facilities. You can read more about it here and here.

Before you feel hesitant in booking an Airbnb in Singapore, fret not! It is not illegal for guests to stay in Airbnb in Singapore for any duration. But the fear of being fined has led to some guests being chased away by condominium security guards. You can read more about it here

Alright enough about laws. We are here to explore the distribution of “risky” listings in Singapore. In this project I will be using the latest Airbnb dataset from Inside Airbnb.

Data Preprocessing

Before we dive into the exploration of features in the dataset, I would like to create a new feature specifically for the context of Singapore’s strict law against short-term staying. This will be a categorical feature that takes the value of either “High risk”, “Medium risk” or “Low risk”. Due to the limitation of available features such as the classification of private properties and HDB properties, the categorical feature is defined by the following:

  • High risk: Minimum nights < 90 days and No. of Reviews < 90th percentile
  • Medium risk: Minimum nights < 90 days and No. of Reviews ≥ 90th percentile
  • Low risk: Minimum nights ≥ 90 days

Apart from adding a new feature, I have also removed prices that are outliers using the standard 1.5x +/- IQR and minimum nights that extends beyond 365 days. Columns not required for this analysis are also removed.

df.head()

df_head

Exploratory Data Analysis

Price Distribution

The distribution of price has barely any association with the riskiness of listings. However it is still interesting to find out the expected range of price of listed properties so we can know what price we are expected to pay for our stay in an average Airbnb in Singapore.

price_distribution

This distribution of the listing price appears to be right-skewed. If we take a closer look at its distribution by region, we will subsequently find out that at least one region does not conform to the overall average price. This can be easily confirmed using a simple non-parametric multiple independent sample test: the Kruskal-Wallis test.

from scipy import stats
stats.kruskal(*[group["price"].values for name, group in df.groupby("neighbourhood_group")])

## KruskalResult(statistic=472.38454622337827, pvalue=6.282024944436425e-101)

price_dist_region

Like any other metropolis, it is not surprising to see a higher distribution of price in the central region as most tourists will hang around the central region with the abundance of shopping districts/bars/attractions/nightlife. The proportion of private property in central region is also a lot higher as compared to any other regions.

Geo-location Distribution of Listings

It is very hard to visualize risky listings by geo-location since they tend to overlap with each other. The populated map below barely serves any purpose in finding the distribution of risky listings in Singapore:

A Tableau-generated distribution of listings by risk rating

But as soon as it is separated by region, it is clear that majority of the listings are densely populated at the central region:

geo_distribution_by_region

Using the beautiful folium library in python, we can roughly observe the number of listings in each region:

import folium
from folium.plugins import FastMarkerCluster
m = folium.Map([1.38255,103.83580],zoom_start=11)
m.add_child(FastMarkerCluster(df[['latitude','longitude']].values.tolist()))
display(m)

geo_distribution_folium

The folium package allows interactive visualization of geo-location of each listing. From here it is very obvious where most of the listings come from, but just how many of these listings are compliant with the law?

Law-Compliant Listings

In this section, we will look at the distribution of listings that are compliant to the law:

distribution_of_reviews

I used number of reviews as the feature of interest because it makes sense to have many different guests stay in a property that have lower minimum nights, which ultimately leads to having a higher number of reviews. From this scatterplot, we can clearly see the distinction of plots distributed at the 90/180/365 night mark. Since we cannot differentiate which listing is a HDB or private property, I will assume listings with at least 90 nights of stay to be law-compliant. At first glance, it appears majority of the listings do not adhere to the rules of short-term stay. Furthermore, certain plots appear to not make sense. Take the 90-night minimum with 285 reviews for instance. How is that possible? After further investigation, it appears that listing has been around before the law kicked in 2017 which accounted for the high review count. Furthermore, that listing only rents out private rooms that will possibly garner higher review count since it will accommodate higher number of unique guests.

Anyway let’s look at the proportion of listings through a doughnut chart:

law_compliant_pie_chart

When I first observed this, I was taken aback by how many hosts are so daring to be listing their property that goes against the law. But then again, since Airbnb do not give out the exact address of listings until booking is confirmed, it poses a challenge for authorities to catch these hosts, especially when majority of properties in Singapore are multi-storey. I will talk more about this at the end of my analysis. But for now, let’s look at the proportion of risky listings.

Risky Listings

Knowing that majority of the listings are concentrated in the central region, let’s look at the distribution of risky listings in each region:

distribution_of_risk_by_region

Again, really no surprise here. A lot of overseas guests just prefer to stay in the vicinity of local attractions which just happens to be clustered at the central region but how does the distribution fare for the types of room?

distribution_of_risk_by_room

Now we are able to see the difference between listings that are compliant with the law as compared to those who aren’t. Low risk listings have a higher proportion of private rooms being rented out as compared to whole apartment. We can see the average rental price of rooms/apartments in Singapore based on region here.

Very high rental prices in Singapore makes it a wiser choice to rent a room instead of a whole apartment. It happens to be one of the driving forces that makes Singapore one of the most expensive countries to live in. The extra fees incurred by Airbnb makes the choice much more obvious. Most apartment rentals are done through proper real estate agent or private agreement between 2 parties for reduced fees.

And if you are wondering which neighbourhood has the highest number of listings I have compiled the top 5 number of listings by neighbourhood below:

distribution_of_risk_neighbourhood

Take Away

While it remains unclear as to the decision for implementing a tight law on short term stay, it seems that majority of private homeowners support the decision of the 3-month accommodation law:

A national survey of more than a thousand private homeowners commissioned by the URA in the second half of 2018 found that the majority of Singaporeans supported the proposed regulatory framework for STA. — CNA

The URA’s decision to maintain the status quo in Singapore is a signal that it recognises the complexity of the issue and needs more time to study it. — CNA

As a resident of a private property and someone who has sublet my spare rooms to guests before the law kicked in, I am able to see both sides of the coin (supportive and unsupportive).

Unsupportive: As compared to HDB homeowners, private homeowners pay a premium for on-site facilities like swimming pools/tennis courts/gyms/private carparks etc.. It will be quite unsightly to see the possibility of these facilities being misused by tourists especially since maintenance of private property facilities are not as detailed as public facilities.

Supportive: With the emergence of modern travelling, Airbnb has always been one of the best platform to provide affordable accommodation with more space given per dollar paid (eg. entire apartment for the price of a hotel room). As a millennial traveler who is conscious about my spending habits, I will always choose Airbnb for its price and convenience whenever I travel overseas. So I see this initiative as a means to target guests who prefer to spend their travelling budget on experience rather than the quality of their accommodation.

Looking at the proportion of listings that are clearly non-compliant with the short term accommodation law, I do wonder about the strictness of the law being enforced. Maybe it’s just not worth hunting down homeowners as long as guests do not cause any disruption? Maybe.

Having a property listed with minimum nights of less than 90 days does not necessarily mean that it is illegal for homeowners to lease out their property for guests.

As I mentioned, when a guest makes a booking through Airbnb, fees will be charged as a form of insurance and accountability for both the homeowners and guests. It is possible that a homeowner may choose to rent their property for a month to “evaluate” their guests before they extend it to 3 months.

This is the exact procedure I went through when I was staying in Canada for my winter exchange for 4 months. All the homeowner has to do is block the remaining 3 months in their Airbnb listing and based on trust we pay our rent by cash. We managed to cut our rental by about $100–200 per month for the entire apartment.

Shoutout to Carl from Montreal if you are reading this! It has been a very pleasant and memorable winter!

Other Loopholes

It is quite common of homeowners to inform their guests to let security know that they are friends coming over to visit. It is not illegal for friends to stay over at another person’s house in Singapore.

This highly depends on the security guards of different apartments. Some will let you off the hook but some will go their way to chase you out like what happened with the poor NZ family.

What should you look out for?

If you are still reading and are willing to book an Airbnb in Singapore despite the risk involved, read on. But please understand that I am merely speaking from what I would do if I were to make a booking and in no way will this guarantee that you will be safe from being evicted, albeit a reduced risk.

  • Check the number of reviews and most recent reviews. If the last good review is just a couple of days from when you visit the site, it means the previous guest was not evicted from the property.
  • Try to avoid listings that are in condominiums, these are places with security guards and they are the ones responsible for kicking you out. If you must stay in one, message your host and ask him/her to fetch you from the guardhouse. It will convince the guard that you are acquaintances. Also do not travel with a huge group, it will only raise suspicion.
  • In any case, always message your host that you are aware of the laws about short-term accommodation in Singapore and seek instructions for your upcoming stay.

So now that you are aware of short term accommodation laws in Singapore, are you going to risk staying in an Airbnb?

I know I would.

If you are keen about replicating my work for your own use, here is a link to my Github repository

Happy Coding!

Bobby Muljono
Data Analyst

Just an average Joe with a passion in data science