Weekly Economic Update 05-23-25: Artificial Intelligence; and Existing Home Sales
Not much data this week...so let's try something new
The views and opinions expressed in this post are solely those of the author and do not necessarily reflect the views of the Georgia Institute of Technology or the Georgia Board of Regents.
There was only one single piece of economic data released this week - existing home sales. That hasn’t happened since I started producing this blog. So, I am going to do something a little different.
Let me preface my comments by saying…..I’m old. I acknowledge that. I’m not as old as some of you, but quite a bit older than many of you. I’m a child of the “personal computer age.” IBM released its first desktop personal computer (PC), the IBM 8180, when I was 11. Monochrome monitor, dual-5 1/4” floppy disk interface, and an Intel 8088 chip, running at the blistering pace of 5 megahertz, and could interact with a staggering 1 megabyte of memory. It was amazing. Microsoft provided the operating system (MS-DOS), and of course, you had to have the Epson dot-matrix printer to complete your setup. It looked just like this:
It retailed for $1,500. A staggering sum, but somehow, my father was able to purchase one. I sat down in front of it for the first time at the age of 11. By the time I was 15, I was an expert at Lotus 1-2-3, and by 18, I was building spreadsheets so sophisticated that they dazzled mere mortals. In fact, IBM hired me at 18 to build financial spreadsheet models for one of its national divisions. In college, I ran circles around my computer science professors.
During this time Video Cassette Recorders (VCRs) were also gaining in popularity. My grandparents certainly didn’t have a PC, and while they did have a VCR, despite their best efforts, they were completely unable to comprehend how to program it so that it would record their shows while they were away. It might as well have been alchemy. It just wasn’t going to happen. They were two generations behind me, but technologically, they might as well have been on another planet. They were very smart people, and I could never understand their lack of comprehension. They had reached their technological limit.
Fast forward a decade, and the World Wide Web (WWW) was introduced to us by the first graphical browser, Netscape. As I had in the past, I embraced this new technology with extreme vigor. I started a company, Home Page Design, Inc. (HPD.net), to build commercial websites. I purchased dozens of popular domain names, realizing they would be the valuable real estate of this new cyber world. My parents and their generation were less impressed. They had embraced the PC, but this Internet thing was “a fad.” Even the famous Nobel Prize-winning economist, Paul Krugman, in 1998, infamously said of the Internet that “it will become clear that the Internet’s impact on the economy has been no greater than the fax machine’s.” Many of that generation now use the Internet daily, but they were slow adopters, and even now, rarely use it to its full potential.
Fast forward another decade, and on January 9, 2007, Steve Jobs introduced us to the iPhone. If the Internet was a fad, this was completely useless. Why would anyone want to have a “smart” phone in their pocket?!? Many in my parents’ generation never bothered to get one, and as recently as a week ago, I was showing someone in that generation how to text a photo. They had reached their technological limit.
And now, almost 20 years later, we have Artificial Intelligence (AI). We have AI assistants (“Chatbots”) like ChatGPT, Claude, Gemini, DeepSeek, Grok, etc. We have video generation and editing tools like Synthesia, Runway, Filmora, and OpusClip. There are image generation tools like GPT-4o and Midjourney, as well as notetakers and meeting assistants like Fathom and Nyota. I could go on, but you get the point. AI is everywhere, and getting better, not exponentially, but “tetrationally.” You may not be familiar with that term. Exponential growth is repeated multiplication (e.g., 2^4, or 2x2x2x2=16). Tetrational growth is repeated exponentiation (e.g., 2^^4 = 2^2^(2^2) = 2^16=65,536).
I was discussing a work problem with my brilliant young team a week or so ago, and one of them said to me, “Just use AI.”
I am going to be very transparent with you…using AI never crosses my mind. It just doesn’t. It isn’t on my radar. It isn’t in my toolbox. I simply do not consider it in my daily routine. I seem to have now reached my own technological limit. And I’m OK with that. I finally realized why my grandparents were never able to program the VCR. I totally understand it now.
I have been very good at what I do for more than 35 years. And I am very good, using the tools I already have. Will AI change the world? No doubt. It could be for good or bad. Hopefully, one day we will use AI for more than making silly baby videos and personal therapy. Could AI make me even better at what I do? Absolutely. But again, I’ll be brutally honest. I have four years left in my professional career. (Well, four years, two months, and eight days, but who’s counting….). And after four years, two months, and eight days, I may very well never open a spreadsheet again. I may never again make another slide-driven presentation. And I am sure I will never build another econometric forecasting model. Frankly, I plan to go fishing with friends, travel the world with my wife, walk on the beach with my dogs, spoil my grandchildren, and take afternoon naps. I don’t need AI for any of that. I’m good.
But that is just me. Of course, the generational limits I describe above are broad generalizations. There are obviously exceptions. There are people much older than I who are all over AI and use it daily. (I’m looking at you Dan!) And that is great. Good for them. I’m just not that interested.
Which finally leads me to this week’s economic update. As I mentioned before, there was only one piece of economic data released this week. I didn’t know how I would make the update long enough. Someone suggested I do a test and let AI do it this week. That was an interesting idea. I thought, “Why not?” So here is what I'm going to do….the first section below is my writing as usual. A couple of paragraphs and a chart. Then, I will do the same data again, but using only AI. (That will probably take me half the night to figure out, but I’ll get it done. I can program a VCR after all. Bet some of you youngsters can’t do that! Of course, you would just ask AI how to do it…)
This could be really interesting. It could also be really bad. The AI version may be much better. You may decide that you don’t need this update anymore. You may find that AI does a better job than I do, and you can just have AI generate an economic update for you each week. I hope that doesn’t happen, but we’ll see how it goes.
Existing Home Sales
Existing home sales fell 0.5% in April to an annual sales rate of only 4 million units. That was far worse than the +2.0% growth that experts were expecting (full release here). Those results made this the worst April since 2009, just after the housing crisis.
The inventory of existing homes for sale rose to 1.48 million units - the highest level since the beginning of COVID.
However, more supply isn’t bringing prices down. The median sales price for an existing home rose 1.8% from last April to $414K - a record for the month of April. However, much of the growth in the “median” price is being driven by sales activity of high-end homes. But at least price growth does seem to be slowing. Another positive sign is that first-time home buyers made up 34% of all purchases, the highest share since July 2020.
Existing Home Sales (AI version)
This morning, the National Association of Realtors (NAR) released the Existing Home Sales data for April 2025, revealing a continued slowdown in the U.S. housing market. Sales of existing homes, which include single-family homes, condos, and co-ops, dropped to a seasonally adjusted annual rate of 4.00 million, a 0.5% decline from March 2025’s 4.02 million and below the forecasted 4.10 million. This marks the weakest April sales pace since 2009, reflecting persistent challenges such as high mortgage rates, averaging 6.99% for a 30-year fixed loan, and record-high home prices. The median sales price rose 1.8% year-over-year to $414,000, the highest April median on record, exacerbating affordability issues for first-time buyers, who comprised 32% of the market, consistent with last year but below the historical 40% average.
Despite the sales decline, inventory levels showed improvement, offering a silver lining. The supply of existing homes for sale increased 20.8% year-over-year to 1.45 million, equivalent to 4.4 months of supply at the current sales pace, the highest April inventory since 2020. This rise in listings, coupled with slower sales, has slightly tempered price growth, with the annual median price increase of 1.8% being the smallest since August 2024. However, economic uncertainties, including concerns over potential tariffs and inflation, continue to weigh on buyer demand, particularly as contracts for these sales were likely signed 1-3 months prior. Analysts note that while increased inventory could ease affordability pressures in the long term, high prices and borrowing costs remain significant barriers, signaling a cautious outlook for the housing market.
Who wins?
OK, so which is better? I like mine, but I am obviously biased. The AI one is pretty good. Seems a little dry, but certainly provided a little more analysis. Interestingly, our numbers don’t always agree. For example, I have new home buyers at 34% of purchases, and AI says 32%. For the most part, the AI summary is very much cut-and-pasted straight from the release. However, the release clearly says, “First-time home buyers were responsible for 34% of sales in April, up from 32% in March and 33% in April 2024,” so I don’t know where the AI’s 32% came from.
So which do you prefer? I guess if I have a lot of people “unsubscribe” this week, I’ll have my answer. If nothing else, the exercise did introduce me to how easy it is to use AI for basic data gathering needs. But the fact that the numbers aren’t consistent with the release does give me pause.
One More Thing…
Earlier this week, I attended the Georgia Economic Developers Association (GEDA) Spring Workshop, which was held at Chateau Elan in Braselton. I was able to interact with several of my readers, and I deeply appreciate their kind words about this update. They were very encouraging, and I am glad to know so many of you enjoy it. I hope you will stick with me and not go with AI!
As always, thank you for subscribing and reading this weekly update. If you find it informative, I invite you to click/scan the QR code below to join as a “member” or to buy a coffee or two (or five) and support this effort.
Have a great Memorial Day weekend! Lots of good data on tap next week!