Welcome back to Home Economics, a newsletter about the American housing market. This is The Week in Review, my weekly recap of macro and real estate news. Articles with a ◎ are free. Those with a ◉ have free previews but are only accessible in full for paying subscribers. Upgrade your subscription here:

🚨 Last week I shared the Housing Policy Analyst, a ChatGPT chatbot. I’ve had some really encouraging feedback from subscribers—and this is just the beginning. The Analyst is trained on a repository of academic papers related to housing, especially affordability, zoning regulations, land-use policy, and the economics of housing.

You can ask the Analyst questions like, How might increasing housing supply affect prices, according to recent studies?

I’ve calibrated the Analyst to ensure it helps users understand the balance between empirical findings and theoretical frameworks. In terms of policy debates on housing issues, I curated the papers to represent the range of views among housing researchers. The Analyst is trained to deliver responses in a nerdy but casual tone readers of this newsletter might appreciate.

Paying subscribers can find the link to the Housing Policy Analyst here ◉.

Next week is a very busy one for housing related economic data. The key event is of course the Fed meeting which takes places over two days, starting on Tuesday and culminating with Powell’s press conference on Wednesday at 2pm.

Expected rate cuts were an important part of my forecast, from last year, that mortgage rates would be below 6% by the end of 2024. We are now spitting distance from that target. I most recently recapped the reasoning behind my forecast here ◉:

The Week in Review Aziz Sunderji · Aug 4 Read full story

In addition the Fed meeting, we also get the NAHB homebuilder survey, home starts, and existing home sales.

Subscribers can download the housing economic data calendar (a Google calendar) here ◉.

News: The Fannie Mae Home Price Sentiment Index (HPSI) rose slightly
Views: Buyer sentiment will not meaningfully improve despite falling mortgage rates

The HBSI inched up but remains at a low level. Overall, the index continues to paint a grim picture for homebuyers—the vast majority think it remains a poor time to buy a home, due to high mortgage rates and high prices. Sellers think it’s a relatively good time to sell.

That said, the marginal changes are towards a better balance: buyers were slightly more optimistic in the latest reading, owing to lower mortgages rates and greater supply. Meanwhile, sellers—especially in the Sunbelt—were less upbeat.

Many housing analysts are expecting further declines in mortgage rates to boost buyer sentiment. I don’t think that’s quite right. Incomes overshadow rates—and a declining proportion of buyers are saying their incomes are higher than a year ago. In fact, the reason rates are falling is precisely because the outlook for household incomes is more dismal.

News: Zillow’s index showed homes prices rising at a sluggish 3% y/y
Views: I expect a prolonged period of tepid price gains like this

Zillow’s index showed has been showing a gradual deceleration in the pace of price gains for months. In the most recent reading, from earlier this week, gains were just 3% y/y. I think these kinds of muted gains, if not even slightly lower, will continue (outright price falls are unlikely: home prices do not fall absent economic or financial calamities, see link below ◉).

Waiting for Home Prices to Fall? Don't Hold Your Breath May 30, 2023 Read full story

For one, incomes are the best predictor of home prices, and they are softening. Second, home prices are now very high versus incomes and rents. A period of consolidation is in order.

Further reading:
Why Zillow is a better index to follow than Case-Shiller, here ◉
Rents are rising rapidly in the biggest cities, like New York, here ◉
Incomes are correlated with prices, here ◉

The new, more pensive, ChatGPT. Earlier this week OpenAI released a new model. Unlike prior upgrades, say from ChatGPT 3 to 4, the new model behaves entirely differently from any earlier one. If ChatGPT 4o—the newest “mainstream” model—behaves like an incredibly smart but impulsive know-it-all, this new model (“o1 preview”) is a more methodical genius. As OpenAI describes it:

We trained these models to spend more time thinking through problems before they respond, much like a person would. Through training, they learn to refine their thinking process, try different strategies, and recognize their mistakes. 

It scores far higher on IQ tests than prior models. At the same time, it’s lousy at some more basic tasks (not unlike real world geniuses!). In any case, I am eager to unleash it on some of my thornier challenges, and excited for what it suggests about the possibilities for AI in the future.

ChatGPT-o1 preview, here ◎.

Apple music. I remember when Spotify first came out. It was 2008. I was living in London, but had just finished grad school in Sweden, and got early access through a Swedish friend. For a music lover, it was mind-blowing. I had been paying $20 for a CD of 20 songs, most of which were invariably duds—and now could pay half of that, for all the songs in the world, in my pocket. I’ve been a Spotify evangelist since, despite the app’s infuriating user interface. But Spotify hasn’t kept up with the times. In particular, you can stream at far higher quality (“lossless audio”, in the parlance) from Apple Music, for the same price. I always support the innovative, scrappier firm (amazingly, Spotify is the scrappy one here), but the difference in offering has grown too large.

Apple Music, on Mac OS and iOS, here ◎.

I hope you’re having a great weekend,

PS: I am in the midst of moving homes, and there has been a dearth of “fun stuff” in my life—send in any recommendations and I will include them next week!