RTB Rent Register

While building my tool to gather data from the RTB rent register, I've had long discussions with ChatGPT about how to code and optimize it.
Now I asked for its thoughts on the rent register itself. The answer is pretty interesting. Even ChatGPT knows it's not a register of market rents.

The final recommendation is fascinating. It suggests I could compute statistically defensible rent ranges, identify regulatory distortion in the data, and present this in a legally defensible way.

Quote below and full response attached.

What the RTB Register Actually Is (and Isn’t)
It is:
  • A registration database
  • Based on actual agreed rents
  • Legally required filings
  • Time-stamped through RT numbers
  • Large and nationwide
It is NOT:
  • An asking-rent database
  • A full census of all rental units
  • A real-time system
  • Cleanly deduplicated
  • A controlled statistical sample
 

Attachments

Last edited:
I had a look at the data for @patfert1 and here is the first finding - quick overview. Very interesting - we can see now why some properties are not showing - there is a huge gap in the database. The same pattern exists for my area, which patfert1 kindly ran for me and explains why my property never comes up even if I use a different Eircode.

The early part of the gap might be due to Covid as few properties changed hands. But 2022 and 2023 should be back to normal. Is it due to when they had problems with their database? Or the fact that this was a period of high inflation so (conspiracy theory alert!) they wanted to leave out (especially new) tenancies that started during high inflation when landlords might have pushed the boundaries of what they could get?

clontarf all rent.webp
donegal all data.webp
dublin 8 apts all.webp
firhouse all data.webp
 
Does it pick up Fora?
Commonly used LLMs can do that.
I've seen ChatGPT and Gemini cite Askaboutmoney in answers to certain questions about Irish personal finance issues.
In any case, the LLM should cite its sources either by default or if asked explicitly.
 
I cannot imagine EAs like Clontarf and Firhouse having only around 200 rental houses each.

If you did a search with BER exempt it would show a lot, is my guess.

But 2022 and 2023 should be back to normal. Is it due to when they had problems with their database?
Are those problems publicised? I can FOI again to see if those dates are included. I'll also ask for clarification on how the score is calculated and whether the data is chopped 10%+/-

Is there anything to be said about the RTB numbers being incorrect for those years? I'm not sure how to explain this, did we nail down what the two numbers are after the RT number? So if its 25 after, is the property just renewed with a new tenant? And with that data you visualised there, I presume it's running off the first numbers, not the ones after.

Fascinating that those years are underrepresented, though I'd say there's no conspiracy that can't be explained away as incompetence.
 
Taking Firhouse 3 bedroom as it has a decent sample size. BER does have an effect on the rent (below). It seems likely that an A or B-rated propert is a newer property than a C or D-rated property. But interestingly B-rated properties command a higher price than A-rated. Maybe closer to transport/shops etc vs newer, less well located estates?

firhouse 3 bed BER.webp


When we take size into account and look at the rent per sqm (below), B-rated definitely win out and A and C rated are equally high rents ie equally attractive, despite A-rated properties being the largest. D-rated properties tend to be smaller, and less efficient so this is probably why they are less desirable. So maybe B-rated are conveniently placed and a decent size and energy efficient. C-rated are probably even more central, and hence smaller, and the BER is still reasonable. Just a hypothesis...

firhouse 3 bed ber by sqm.webp
 
Are those problems publicised? I can FOI again to see if those dates are included. I'll also ask for clarification on how the score is calculated and whether the data is chopped 10%+/-
I haven't seen this gap publicised. FOI would be good. And clarification on the scores etc. e.g. on their website they don't give location as one of the criteria they use for matching, but as @patfert1 pointed out, different eircodes in the same electoral district give different results.
 
I was just listening to today's RTB seminar on rent setting. In the Q&A the RTB said that the register doesn't include cost rentals.
Posters on here have identified cost rentals close to their own rentals coming up on the searches.

Some searches by posters here have, from memory, shown up a number of units of same size and BER all at the same low rent - have to be cost rental.

Also quote from a Business Post article by Ian Guider

In one example, in Donaghmede, Dublin 13, checking comparables for a two-bedroom A2-rated apartment of 78 square metres in a build-to-rent (BTR) scheme, that is currently renting for €2,713 per month shows ten RTB examples, all recorded at exactly €1,635 per month. That represents a gap of roughly 40 per cent between the current rent in the scheme and the level implied by the RTB register. The database entry for each comparable also records zero bed spaces, an obvious error, highlighting potential data-quality issues.

Those 10 must be cost rentals.

Unless they have stripped out cost rentals going forward, the RTB are lying or don't realise that cost rentals are showing up (either is entirely possible)
 
If you did a search with BER exempt it would show a lot, is my guess.
I'll re-run it tonight, for the clontarf and firhouse data. I didn't include BER exempt in any of the input criteria for the tests I ran, so it will be interesting to see if adding it has an effect. Having said that, the Firhouse and especially Clontarf results I got already include many unrated houses.

I can FOI again to see if those dates are included. I'll also ask for clarification on how the score is calculated and whether the data is chopped 10%+/-
I suggest putting it in more general terms, is there any filtering criteria used to remove any data, such as top or lower x%. Also ask explicitly if cost rentals included. Also ask if the address is used in step 2 search, because empirical evidence suggests it is.

Is there anything to be said about the RTB numbers being incorrect for those years? I'm not sure how to explain this, did we nail down what the two numbers are after the RT number?
The numbers after are almost certainly when a tenancy was last reviewed, since they are very common and all 2025 or 2026.
The first 4 digits after RT are most likely when the current tenancy started, but we cannot distinguish tenancies when a new tenant moved into a property which was on a previous RPZ tenancy for some years.
B-rated properties command a higher price than A-rated
My input criteria for Firhouse and Clontarf houses went up only to B3. Having said that there are a large number of Ax-rated in the results, especially for Firhouse When I re-run including BER exempt I will also ask for A-rated.
 
I can FOI again to see if those dates are included. I'll also ask for clarification on how the score is calculated and whether the data is chopped 10%+/-
Could you also ask:
* What is the rationale to show only 10 entries deemed most suitable by an arbitrary definition of match-score?
* What is the reason to not allow a full view into ALL potentially relevant properties?
The above two are particularly necessary for landlords in areas where sparse numbers of rental properties are meaning returned results have low match scores (<60%).
* What is the reason to not provide additional necessary information to allow a landlord to make a properly informed decision about rent level, specifically: How long the property has been under RPZ rules; when did the current tenancy start (+ anything else you can think of)
* What does number of bed spaces mean and what does it signify if it is zero?
* Why is there no information about the standard of the property (standard of decoration, luxury equipment, garden area, ...), again which would be necessary for a landlord to make an informed decision.
 
Why is there no information about the standard of the property (standard of decoration, luxury equipment, garden area, ...), again which would be necessary for a landlord to make an informed decision.
Perhaps consider balancing that by replacing "landlord" with "landlord and tenant".

Their reason, should they answer, would probably be - that they have no idea what state of the properties on the register are in. There is a straw to grasp in the BER rating which might indicate a property is quite new.

They know nothing about how old, cheap, nasty and moldy bathrooms are, how smelly & grubby the carpets, how basic the kitchen, how depressing or uplifting the experience of standing in the property. They know nothing about the quality - not even if there's parking - and to suggest these things are not relevant to establishing a fair market rent is absurd.

There is one group of landlords who will be pleased with these new rent rules and the pressure to price in line with the RTB rent register - slumlords. They can push up the rents on their kips to be at or just less than the highest of the RTB's selection of 10 and the RTB's IT system will never flag them as charging more than a fair price, and any complaint by a tenant will be dismissed since the slumlord is being fair by the RTB's own definition.

How long the property has been under RPZ rules; when did the current tenancy start (+ anything else you can think of)
I think "when did the tenancy start" is indicated within the RT number, no? But is in any case a bit of a distraction - any new tenancy of a property previously under RPZ rules is no more relevant than an old one because the same restrictions on rent level have applied. It's important that the focus is shifted on to what matters, which is the first part of your sentence.

"How long the property has been under RPZ rules?" - or asked a different way

"when was the property first introduced the the rental market, or reintroduced after an absence of more than two years?"

...
as these would indicate a property for which the rent may reflect recent actual market rent.

They would probably obfuscate in a reply to that with something vague about only being able to look at data - so a data sympathetic translation for them might be something along the lines of

"the latest date of when each property first appeared on the register, or reappeared after an absence of two years or more"

Their data will be a bit unreliable especially going back several years (eg landlord has property become vacant, leaves vacant for two years perhaps while renovating, but forgot to, or was unable to, end the registration of the previous tenancy) but even with some errors that would be a vast improvement in the relevancy of the data used by landlords and tenants to 'discover market price' - and it is in any case the more recent, cleaner (because of annual re-registration), data that is most relevant.
 
They would probably obfuscate in a reply to that with something vague about only being able to look at data - so a data sympathetic translation for them might be something along the lines of

"the latest date of when each property first appeared on the register, or reappeared after an absence of two years or more"
"the latest date of when each property first .." doesn't make sense as first is by definition earliest not latest.
Also not everywhere became an RPZ at the same time, so better wording might be

"the date of commencement of the first tenancy at that premises that was subject to an RPZ, or the date of commencement of the first tenancy at that premises after an absence of more than two years that was subject to an RPZ at its commencment"
 
There is one group of landlords who will be pleased with these new rent rules and the pressure to price in line with the RTB rent register - slumlords. They can push up the rents on their kips to be at or just less than the highest of the RTB's selection of 10 and the RTB's IT system will never flag them as charging more than a fair price, and any complaint by a tenant will be dismissed since the slumlord is being fair by the RTB's own definition.
So true. Another reason why the biggest loser in all of this charade will be the renter..
 
Back
Top