A rent increase letter should show its working

Every year my company sends about 52,000 letters asking tenants to agree to a higher rent. German law lets a landlord raise the rent of an existing tenancy towards the rent customary in the locality for comparable flats, within limits on how often and by how much, and only with the tenant's consent. The letter has to give reasons. Where a town publishes a rent index that covers the flat, the reason is usually the index. Where it does not, the landlord may instead name at least three comparable flats whose rents are as high as the rent requested.

That last route is lawful, and it is where I think large landlords have most to answer for. A company with thousands of flats in a town has an enormous choice of comparisons. Its tenant has none. The tenant receives three addresses, three rents and three floor areas, cannot see inside those flats, and cannot see the hundreds of others the landlord could have named. The asymmetry is built into the method. Whether it becomes unfair depends entirely on how the landlord chooses.

Three flats chosen by the landlord

I know how the choice can go wrong because I built a system that made it badly. In 2018, as head of rent and portfolio analytics at Haselwerder Wohnen, I led the team that automated our comparison-flat letters. The system filtered our portfolio for flats of similar construction period, size and fit-out in the same town, and then picked three. Its instruction was to support the highest increase the law allowed, so it picked the three highest rents among the eligible flats.

Nothing in that design was illegal, and every flat it named was comparable by the criteria we used. The effect was still predictable. Our comparison sets came from the top of the local distribution, often recently relet flats whose rents reflected a tighter market than the one the sitting tenant had signed into. By 2021, 19 per cent of tenants who received such a letter objected or refused consent, against 8 per cent where we relied on a rent index. We were suing about 640 tenants a year for consent. A regional tenants' association noticed that our comparison flats tended to come from the same few buildings, and it was right.

When I became chief data officer, I reopened the system I had designed. The question I put to our management board was not whether our letters were lawful. They were. It was whether a reasonable tenant, shown how we chose, would accept that the choice was fair. I did not think one would.

What a fair comparison looks like

A fair comparison set, in my view, has three properties. It is chosen by a rule a tenant could apply. It sits near the middle of what is comparable, not at its upper edge. And the tenant can see enough to check it.

We now select the eligible flats nearest to the tenant's own on stated characteristics: the same district, the same construction-year band, floor area within a fifth either way, a comparable standard of bathroom and heating, and a similar position in the building. Among those we choose the flats closest to the median rent of the comparable set rather than the highest. Since 2023 every comparison letter has carried a one-page summary showing how many comparable flats we hold in that district, the range of their rents per square metre and where the requested rent falls within it.

The cost was real, and I argued for the change knowing that. Our chief operating officer, who answers for rental income, opposed it in the management board. He pointed out that the law did not require it, that competitors would carry on choosing as before, and that we would give up roughly €6 million a year of increases we were entitled to request. He was right on all three points. The board accepted the change because the old method was costing us more in disputes and goodwill than it earned, and because we did not want our rent letters to depend on tenants being unable to check them.

The average increase requested in comparison-flat letters fell from 9.6 per cent to 6.9 per cent. Consent without dispute rose from 81 per cent in 2021 to 93 per cent in 2025, and actions for consent fell to about 140 a year. Increases now take effect about four months sooner on average, because fewer of them wait for a court. Even after those gains we collect less rent than the old method would have produced. I regard that as the price of a method we can defend in public.

Publish the rule, not the model

Some colleagues in the sector argue that the answer is better modelling: hedonic regressions that price every attribute of a flat and estimate its customary rent precisely. I spent my first six working years building qualified rent indices with exactly those methods, and I respect them. But a model does not solve the tenant's problem. A tenant cannot check a regression. A tenant can check a rule.

That is why I argue for publishing the selection rule rather than a scoring model. Ours fits on half a page. It tells a tenant which characteristics we compare, how wide each band is and how we choose among eligible flats. A tenants' association can test it against its members' letters, and one has. When it found that our construction-year bands treated buildings modernised in the 1990s as if they had never been touched, we changed the bands. That correction would not have happened if the method had lived inside a model only my team could read.

A published rule also disciplines the landlord. When a rent campaign falls short of plan, the temptation is to adjust the selection quietly. A rule in the open makes any adjustment visible, so it has to be argued for.

Rent indices deserve the same treatment. A 2021 reform obliged larger German towns to publish rent indices and strengthened the surveys behind qualified ones. Large landlords supply a significant share of the rents in those surveys, which gives us influence over an instrument that later limits what we may ask. I think that influence should come with obligations: deliver complete rather than selected data, tell the town which of our flats were recently relet, and be open about how we answered.

None of this makes a rent increase welcome. Tenants are being asked to pay more for the same flat, and many are doing so after several years of higher heating bills. What a landlord controls is whether the request can be understood and checked. A letter that shows its working will still be resented by some readers. It will be disputed by far fewer, and it will be fair to the tenant who cannot see the other flats.