MultiBlog1 | vfasite

Breaking the Cycle: Projecting Grape Prices into an Ambiguous Future

I was recently talking with a colleague about where grape prices are going and he expressed his belief that some winegrapes’ prices are no longer subject to the 14 year cycle (from crest to crest) that has historically determined grape prices. I hear this a great deal in various forms, but I also hear from people who have been in this industry for a long time that this belief spreads during every upward swing time and it is always proven wrong. I thought it would be interesting to research this, so I’ve been charting grape prices across various regions and varieties. Most still conform to the historical, cyclical pattern, although the cycle is probably more like 12 or 13 years from crest to crest, not 14.

Unsurprisingly, Napa cabernet has clearly broken out of this model. Below is a chart of Napa cab prices, normalized to 2013 prices, meaning that it has had the effects of inflation removed. The y-axis is the price per ton in 1991 prices and the x-axis is the number of years since 1990.

As you can see, Napa cab was due to fall of the cliff in the early 2000s, but instead plateaued somewhere south of its peak, started to rise a bit, fell a bit during the Great Recession and is now slowly but steadily increasing. The only question is what hybrid we have of flat pricing, an upward trendline and a cyclical effect. My model for forward projections takes all of these possibilities into account, weighting them according to past correlations. This is a somewhat simplistic way of estimating prices, but the differences between different methods create only minor differences in output, as evidenced my VFA’s empirical testing. Of course, a model can always be built to someone’s specific assumptions and, more importantly, should be custom built for each vineyard. As my readers may have read in an earlier blog, I expect a pretty significant jump in Napa cab prices for next year, equal to over $300, a rise of about 5%, which hints that we are seeing both an upward cyclical effect and a general upward trend.

Due to regulations and lack of room for more vineyards, Napa cab has hit a saturation point where it is unlikely that planting trends will ever again create a long market. This, of course, has a strong effect on how to do long-term modeling for Napa cabernet, but the new model requires only moderate tweaking to produce reliable results. What about varieties/counties that are in a state of only partial saturation? These will likely present a greater challenge, due to greater ambiguity. I study Sonoma County chardonnay pretty intensely, so I thought that would be a great example to look at on the blog.

The chart below is similar to the one above, but includes a forward projection to the right of the black line, using a generic model that would be my best fit if I could use only one model for all grape/region combinations. Note that these are actually prices for all of District 3 and prices are in constant 2013 dollars (I’m using whatever charts I already have, sorry to have not standardized them, but the nominal price is not the point anyways.)

Three things jump out at me right away:

My preliminary research shows that most winegrapes had prices knocked off by only about 8% by the Great Recession, but in the case of Sonoma County chardonnay, it may be much more, as it looks to have shortened the upward part of the cyclical trend.
Visually, one is led to suspect that, like with Napa cabernet, Sonoma Chardonnay may be headed into a new model, where it has only a slight upward trend and little cyclical movement.
If the above point is correct, then the forward projection is an anachronistic error. This would be unfortunate for Sonoma chardonnay growers, as this model predicts sharp prices increases in the next couple of years.

So, what do we do to properly predict where Sonoma chardonnay is going? Three things (yeah, three seems to be the magic number in this post):

As with all projections, VFA builds out a model specific to each situation.
This is a situation that calls for scenario analysis, where we would use one model based closely off of historical data for Sonoma chardonnay (similar to the chart above) and then a second model that would better model trends similar to those seen in Napa cabernet. We could then assign probabilities to each and modify both the probabilities and the two models on an annual basis, based off of observations and statistical analysis, until we have settled on one. Though the process has some complexities, the output is rather easy to apply to the real world.
In at least one and maybe both of the models, we would use sensitivity analysis. This helps us understand the range of possible outcomes, which allows us to use different inputs for different situations. For instance, we may use a median case when applying for a loan, but use a 5th percentile case for determining whether or not to take out the loan package in the first place. That is, we can use the average case when talking to the bank, but demand a 95% probability that the vineyard can support the loan package. Finally, the sensitivity analysis can also be used to identify which variables and influences are most critical to achieving the revenue per ton that is needed. The chart below is the same as the one above, but also includes lines that show the range of outcomes. The green line is the average outcome; the yellow line shows the 20th percentile case; the red line shows the 5th percentile case; the light blue is the 80th percentile; and the darker blue is the 95th percentile. That means that you can use the red line to project forward when you need to know that there is a 95% chance that you’ll achieve that revenue per ton or better.

One final note: this type of modeling is considerably different than next-year projections and is generalized. They are being used only as examples, not to build pro forma financials for an actual situation. Therefore, they would not agree with the projections on this blog for 2014 prices, even if they were put on the same year’s dollar.