[_delirium]

If I'm reading apartmentlist's methodology correctly, they're tracking average prices for apartments that were vacant and have newly been rented (what they call "transacted rent prices"). But they don't track prices for existing leases to long-term tenants, which are typically much more stable than newly turned over units. Depending on the market, that can significantly overestimate rent increases in the broader market. For example, in NYC, about 50% of renters are in rent-controlled units, but they account for a relatively small percentage of new listings, because the units don't turn over often. So you'd get different answers if you average rent for newly turned over units vs. average rent for all current tenants.

(I'm not in NYC, but also in that category: I've rented my place for years, and my rent has increased 0.0% this year.)

[adam_arthur]

Yes, exactly. Why would you ever want to weight in rent controlled units?

The much more useful number is, what is the cost of living for somebody plopped into the economy. Because eventually almost everybody ends up moving and bearing that cost.

Should our price data be forward looking or backwards looking? Obviously from a policy perspective having forward looking pricing data is much more useful and relevant.

Including rent controlled units and existing leases is a methodological flaw of OER. You're right though, if you game the measurement and design the methodology just right, you can produce lower numbers.

If the CPI formula were unchanged from the 70s, we would have roughly equivalent inflation numbers now that we had then.

[_delirium]

For apartmentlist's purposes it makes sense they do it that way, because their target audience is presumably people who are looking to rent a new unit, and those people care about what newly rented units go for. But if we're looking at changes in cost of living in the broader economy, most people don't move constantly, so it doesn't make any sense to exclude the cost of living of people who are staying in the same unit. You end up massively exaggerating both increases and decreases in cost of living, e.g. apartmentlist showed huge deflation in rents in my city during 2020, even though most people didn't move and most people's actual cost of living didn't decrease. But there were a lot of vacant units listed cheaply as people left for the suburbs during covid. Now the vacant units are back up again as people come back into the city, but my rent still hasn't changed either way, and neither has most people's. So you could tell a story of massive deflation followed by massive inflation, but that is not in fact what most people who actually live in the city experienced. So for an overall cost of living number I'd much prefer just an average of all rents.

[adam_arthur]

So you prefer backward looking to forward looking metrics then. That's not useful from a policy perspective.

The Fed is meant to head off inflationary pressures. If they take a full year to materialize in the data, that's a flawed metric IMO. A 20% market rate rent increase will almost by definition feed into a 20% rent increase for everybody in the longer run.

The market rate rent will dictate rents for all over the longer run. Yes, maybe 10% of the population got a good deal on their renewal or whatever. Including that information is not useful or helpful from a policy perspective.

At the very least, they should produce two figures. Market rate CPI and existing tenant CPI.