Relationship Between Ex-Vessel Price & Market Factors

Originally filed by Andy Wink on July 21, 2020, material pertaining to 2020 ex-vessel price projections were updated on November 17, 2020 in order to utilize actual/newer available data and replace projections based on estimated data.

Our members often ask how market factors further down the supply chain impact ex-vessel prices and value. Macro-analysis about supply and demand can be interesting for fishermen, but to steal an English idiom: what’s that got to do with the price of fish? A fair question certainly, but the reality is that fish prices are the result of a broad set of ever-changing conditions which also appear to ebb and flow in terms of importance as time goes by. The challenge of distilling such a complex web of interactions down to a discreet number has always seemed daunting, if not impossible. So, the best that can be done is to provide transparency regarding wholesale or retail pricing changes (higher is better), the strength of the U.S. dollar (weaker is better), and supply fluctuations. In many cases, analyzing these factors really do provide insight about whether ex-vessel prices are likely to rise or fall.

Still, might there be a way to quantify how these market factors affect fishermen’s prices, at least looking backwards? Having a good deal of experience observing market changes and learning the strong/weak points of various data sets, I set out to at least try to find an equation that could provide some answers. Turns out, there are some strong mathematical connections to be found.

First, available data that was likely to have an impact on ex-vessel sockeye prices for Bristol Bay fishermen from a supply or demand perspective was compiled:

- Bristol Bay sockeye harvest volume (supply)
- Global sockeye harvest volume (supply)
- Farmed salmon prices (demand)
- Net processing revenue (NPR) from the prior year (demand)
- Pre-season export prices of frozen H&G sockeye (demand)
- Subsequent first wholesale prices (demand)
- U.S. Dollar Index (demand)
Note: Net processing revenue (NPR) refers to the total first wholesale revenue earned by processors by selling Bristol Bay sockeye less what was paid to fishermen for raw sockeye product.

Next, a series of multiple regression analyses was performed on (average) final ex-vessel values (and base ex-vessel values, though the final ex-vessel price model produced the best fit). Interestingly, not all these market factors had a statistically significant impact on ex-vessel price from 2011 to 2019. In fact, the U.S. dollar index, subsequent first wholesale prices, global sockeye harvest volume, and farmed salmon prices (4/7) did not have a clear impact on final ex-vessel price. Logically, these factors do matter. Still, in some years with a relatively stronger dollar, or low farmed salmon prices, or high global sockeye harvests, the Bay’s final ex-vessel price was still higher or vice versa, producing a low p-value in the regression analysis (meaning the variables were not statistically significant). However, three variables were highly significant predictors of final ex-vessel price and the multiple regression analysis produces a near perfect fit with final ex-vessel prices. Those factors (i.e. statistically significant independent variables) were:

- Bristol Bay harvest volume (supply)
- Net processing revenue from the prior year (demand)
- Pre-season export prices of frozen H&G sockeye (demand)

The data and regression analysis output for the best-fitting model are as follows:

The first thing that jumps out is how closely estimated ex-vessel prices generated by the model align with (average) final prices published by ADF&G. All but one year (2014) is below 3% in absolute difference between estimated and actual prices. It’s a very close fit.  

The R-squared figures are also very high. R-squared represents the proportion of variance for a dependent variable (ex-vessel price) explained by the independent variables. The three independent variables explain 99% of the final ex-vessel price variance and the adjusted R-squared is nearly as high, which simply means the high R-squared is not being achieved simply by adding more independent variables that do not improve the model’s fit.

The Standard Error of the Estimate (SSE) is used to measure the accuracy of the model-generated estimate. The SSE is pretty low, with a 95% chance that actual (past) prices would fall within a fairly tight range of +/- $0.06 of the estimated value. Finally, each independent variable has p-values well below 0.05 indicating they are highly significant and there is a relationship between the dependent variable (ex-vessel price) and independent variables.  

Another interesting thing about the statistically significant independent variables is that they are numbers which are generally known or can be reasonably estimated either prior to the season or before the season is over. Applying the regression formula to a set of estimated independent variables as of mid-July 2020, resulted in a projected final ex-vessel price of $1.43/lb for Bristol Bay sockeye. However, the actual 2020 points for harvest volume and the previous season’s NPR varied significantly from the July estimates.

The actual independent variables for 2020 are as follows:

(a) BB Sockeye Harvest Volume (in Millions of fish): 39.46
(b) HY2020 Net Processing Revenue ($Millions): $208.7
(c) January-May 2020 Frozen H&G Sockeye Price/lb: $3.02

Applying these data to the regression formula, produces a final 2020 ex-vessel price projection of $1.25/lb. See calculations below.  

y = (-0.0176*39.46) + (0.0042*208.7) + (0.4349*3.02) - 0.2442 = $1.25)
y (average final ex-vessel price) = $1.25/lb
The model predicts there is a 95% chance that final ex-vessel price will be $1.25 ± $0.06

Past data can also provide answers regarding the relationship between base ex-vessel price and final ex-vessel price.

The analysis of base/final ex-vessel prices is relatively simple. From 2016-2019, the average difference between base and final ex-vessel price is $0.26/lb. This period generally reflects the current framework of quality incentives, but also includes “adjustment” payments – particularly in 2017 and 2018. The differences in 2019 and 2016 might represent better estimates of actual difference between base and final ex-vessel price due to quality bonuses.

However, it is important to understand that these figures (and the differences) are averages. In practice, fishermen who chill their fish and meet other quality requirements, or those who qualify for production bonuses receive a higher price than those who do not. In 2018, some fishermen received upwards of $1.90/lb. but the average across the fleet was reported to be $1.60/lb. Although base prices in Bristol Bay are typically the same from processor to processor it may be possible to see some deviation there too, in which case the base price would also represent an average.

Therefore, if the model-predicted final 2020 ex-vessel price/lb. is $1.19 to $1.31 (midpoint = $1.25), the expected base price/lb. would be approximately $0.99 to $1.11 per pound (midpoint = $1.05).

Obviously, there are numerous caveats which should accompany using these past market relationships to gain insights about the future price of fish. The model shown above is an explicit effort to test how past market factors affect past ex-vessel prices. There is no guarantee that these relationships will remain static. However, the precise fit of the data does make it a useful exercise.

The 2020 season is already an outlier in two important ways: COVID and poor salmon runs outside of Bristol Bay. The COVID pandemic has had substantial impacts on financial markets, risk, and currency valuations. Although historically the value of the U.S. dollar has not shown a statistically significant relationship with ex-vessel price, that’s not to say it couldn’t happen. These factors could still have a suppressive effect on ex-vessel price in 2020.

Secondly, although global harvest volumes did not appear to have much effect on ex-vessel prices, there are a few things at work here which could make 2020 different. Bristol Bay’s 2020 harvest may very well account for the largest percentage of global sockeye harvest ever, or at least in a very long time. The model decreases expected ex-vessel price as Bristol Bay harvest volume goes up. However, this year’s relatively robust harvest in the Bay will probably not coincide with a large global harvest as it did in 2016-2019. Plus, ex-vessel prices were relatively strong in 2017-2019 despite large global harvest volume. This was probably aided by strong demand for all salmon species, but it is likely that marketing and quality investments played a substantial role in preserving demand at higher supply levels. The clear outcome in recent years is that the supply chain moved large volumes of sockeye at relatively high prices. More research is warranted on that topic. In any case, these facts would provide support for higher ex-vessel prices due to an expected year of short supply (for sockeye). However, if processors (i.e. buyers) suffer losses in other operating segments, it could depress financial resources available to pay for fish in Bristol Bay even if consumer demand is strong.

Finally, this model is not able to provide any insight on what a “fair market” ex-vessel price should have been in prior years, it only fits the statistically significant market factors to the actual final ex-vessel price. The past market relationships are baked into any predicted ex-vessel price values generated by the regression model. The concept of a “fair market” price is highly subjective, but objectively quantifying what the maximum ex-vessel could be would require a different and more exhaustive analysis, as many factors come into play.

BBRSDA invests member funding heavily in marketing, quality, and sustainability efforts that seek to improve the value of Bristol Bay sockeye. The market research contained above is an important part of understanding how downstream or uncontrollable market factors affect fishery value.