Update (4:43p, 4/18/19): The Lake County Winegrowers Association e-mailed me with some more information:
"Thanks to a near-perfect growing season in 2018, the crop was big everywhere in California, including in Lake County. 2018 prices varied slightly in comparison to the previous years but considering the extent of the fire, the prices recorded in the grape crush report reflect both a commitment to quality from the growers and a good collaboration between growers and winery buyers."
Update (12:33p, 4/18/19): Since I first posted this, I got some feedback and, as often occurs, when one is just looking at numbers and not in the trenches understanding all the relevant events that make up reality, it's time for a face palm. Take my analysis with a grain of salt. At least one broker is of the opinion that much of, if not all the unexpectedly low prices is explained by the fires, with a great deal of fruit being sold at lower prices outside of regular contract terms. The biggest buyers, faced with a large harvest, may have been stringently applying contract terms regarding possible smoke taint.
If that's the case, it is good news on two fronts. First off, Lake County isn't nearly as much of a harbinger of future falling prices for a broader geography. Second, my forecasting isn't necessarily off on this one. I would never claim to predict "acts of God."
*** Original Article Follows ***
One area I've wanted to work in more is Lake County. It's beautiful up there, with great land. For that reason, among others, I have made public forecasts of Lake County grape prices. Some of you may have seen presentations of mine recently or had lunch with me, where I've mentioned that Lake County Cabernet Sauvignon prices rose by less than the rate of inflation in 2018 (compared to 2017). I've mentioned that as an indicator that the overall market is softening for coastal wine grapes.
At the beginning of this year, I produced a forecast for the average price for all wine grapes from Lake County. I predicted that Lake County would continue with most other coastal regions in the trajectory of their prices, with maybe a bit more softening: that is, that it would see a small price increase this year. I was wrong. And by a lot.
I predicted a price of $1,793.98. Prices actually took a big hit (9%), falling from $1,755.62 to $1,598.46. This is well within the 60% prediction interval. I predicted that there was a 60% chance that prices would be between $1,556.50 and $2,042.52. That was a wide prediction interval.
Why such a wide prediction interval? Well, for one, I had a lot of questions. Basically, I couldn't tell if Lake County prices had already started falling or if they would continue to crawl along until the rest of the market started to fall, too. I was, probably, a bit too attached to my broader, coastal models that indicated prices wouldn't start falling, in general, until 2019.
The other reason for the wide prediction intervals was that I have found Lake County prices difficult to predict. This forecast was my second worst of all time, of any time horizon, with an error of about 12%. My worst forecast ever had an error of 17%. It was also for Lake County. For comparison, my third worst, one-year forecast ever was 6% and my one year, average error, through 2018, is below 3%.
I intentionally decided to grade this forecast right after Washington. Some regions are easy to forecast, such as Washington, Oregon and the Sierra Foothills and their sub-regions. Yesterday I took a victory lap for a bullseye prediction for Washington grape prices. I decided to follow up with my most challenging forecast. I was really hoping to have made more progress with Lake County prices. But you can't win them all and it's important to share and analyze failures. For one, I think it is important to building trust that I am transparent. Second, it's the only way to get better. (After all, 12% is better than 17%, right?)
Bigger picture takeaway: this is a canary in the coal mine. My regular readers and my colleagues have heard it again and again: I've been predicting since 2012 that prices will dip for coastal regions in 2019. Though I clearly didn't think Lake County prices were softening to the extent they have, I have cited their weakening prices as a harbinger of rough times ahead. Mendocino also saw a (much smaller) dip in prices.
Additional Thoughts (850a, 4/18/19): I've been looking at my models and confirmed something I felt. Yes, I was off by 12%, which is a huge error for me. But, my models indicated a 95% chance of a fall of at least 8% from my 2018 prediction, in 2019. So my consolation is that I could also say that I was off by only one year, depending what happens next year. How this influences my next Lake County forecast (if I produce one) is unclear until I dive deep into my models again.