54 Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ THE HOPS SURPLAS INFLUENCE ON GLOBAL MARKET EVENTS Douglas MACKINNON 1 and Martin PAVLOVIČ 2 Original scientific paper / Izvirni znanstveni članek Received / Prispelo: 31. 10. 2023 Accepted / Sprejeto: 21. 11. 2023 Abstract The U.S. hop industry data measuring the change of inventory and acreage responsiveness to price demonstrated a delayed, reduced, or total lack of, responsiveness in the change of direction of acreage and inventory in response to directional changes in season average price during free market periods. This reaction was referred to as the delayed surplus response (DSR). The data also demonstrated the absence of the DSR during periods in which proprietary varieties reached greater than 50% of U.S. acreage and production. Patented plant varieties offer a legal monopoly over that intellectual property. The absence of the DSR during periods in which a majority of U.S. acreage and production were proprietary indicate a strong degree of control by the owners of proprietary hop varieties over supply and therefore the ability to manage price at desired levels. Keywords: hop market, brewing industry, surplus, Bayesian inference, proprietary varieties VPLIV PRESEŽKOV HMELJA NA GLOBALNA TRŽNA DOGAJANJA Izvleček Podatki o hmelju v ZDA, ki merijo spremembo zalog in odzivnosti spremembe površin na ceno, so pokazali zapoznelo, zmanjšano ali popolno pomanjkanje odzivnosti pri spremembi površin in zalog kot odgovor na usmerjene spremembe sezonske povprečne cene v obdobjih prostega trga. Ta pojav obravnavamo kot zakasneli odziv tržnih presežkov (DSR). Podatki so tudi pokazali odsotnost DSR v obdobjih, ko so lastniške sorte dosegle več kot 50 % površine in proizvodnje v ZDA. Patentirane sorte hmelja nudijo pravni monopol nad to intelektualno lastnino. Odsotnost DSR v obdobjih, ko je bila večina površin in pridelave v ZDA lastniška, kaže na močno stopnjo nadzora lastnikov lastniških sort hmelja nad ponudbo in s tem na sposobnost upravljanja cen na želenih ravneh. Ključne besede: hmeljski trg, pivovarstvo, presežki, Bayesov sklep, lastniške sorte 1 PhD, MacKinnon Report. Hoppy, balanced and unfiltered, USA, e-mail: doug@demackinnon.com 2 Prof. dr., Slovenian Institute of Hop Research and Brewing / University of Maribor, Faculty of Agriculture and Life Sciences, Slovenia, e-mail: martin.pavlovic@ihps.si Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ 55 1 INTRODUCTION Hops (Humulus lupulus L.) are a perennial that grow bines requiring a trellis system capable of supporting the weight of the vines themselves as well as the many cones each bine may produce. Commercial production takes place in latitudes greater than 35 degrees in both the Northern and Southern Hemispheres due to strong photoperiodism requirements for flowering (Henning et al. 2015). In the United States in 2021 we noted 41% of the global hop production with the average hop yield of 2,1 metric tons per hectare (IHGC 2021). This weight represented only the cones, not the weight of the bines themselves. Breweries are the primary customer requiring hops in significant quantities and are responsible for 98% of global hop sales. The demand for other products, therefore, does not affect the hop market in a meaningful way (DHWV 2012, Pavlovič et al. 2022). The hop industry is relatively small. Therefore, hops are not publicly traded. Instead, they are traded in what is referred to as Over-The-Counter (OTC) deals. The practice of OTC trading is an archaic, yet effective, system requiring personal communications between the interested parties, but lacks transparency. The hop industry is therefore an opaque market oscillating between the influences of a merchant oligopoly and a brewer oligopsony depending upon the supply situation (MacKinnon and Pavlovič 2021). Folwell et al. (1985) first introduced the concept of a "lagged supply response" when describing the hop market disequilibrium. That term however does not adequately describe the situation. Supply is highly elastic and responsive to increasing prices. It is the industry's response to declining demand that is lagging leading to the term "Delayed Surplus Response" (DSR) (MacKinnon and Pavlovič 2021). The aggregate and regular methods used by the United States Department of Agriculture (USDA) to gather and calculate data regarding the U.S. hop industry, their policies regarding reporting confidentiality and their consistency for over 100 years makes USDA data removed potential biases and was well suited to serve as Bayesian priors (Bayarri and Berger 2004). USDA season average price acreage and production data demonstrated that in response to deficit situations in 1980, 2007 and 2008, U.S. grower responses were rapid (USDA 2020). 2 MATERIAL AND METHODS 2.1 Calculating the Depletion Rate The U.S. hop stock figures represented inventory in warehouses located in the United States. To achieve a more accurate picture of the amount of U.S. and foreign hops stored in the U.S. during the period analyzed, import and export figures were added to the equation. 56 Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ The depletion rate is the rate of change from one data point to the next similar data point. We measured the depletion rate between annual September 1 hop stocks figures. Alternate depletion rates (i.e., between September year n-1 and March year n or between March year n-1 and March year n) may be calculated and provide some value. For the purposes of this research however the September 1 depletion rate was used and will be referred to henceforth as "the depletion rate". We calculated the depletion rate by taking the September 1 U.S. Hop Stocks value for the previous year n-1, ( ) adding in the total production of the U.S. crop (accounting for processing loss ) for year n-1, (′ ) , and subtracted the September 1 U.S. hop stocks value for year n, ( ). To account for the quantity of hop production lost during processing for any year n into pellets or extract, referred to as (′ ) , we estimated that 97% of the original raw production volume (hops in bale form). Furthermore, we estimated that 3% of the crop remained in bale form. For these hops, it can be assumed there was zero loss. These conversion rates came from a priori knowledge and will certainly change in the future as brewer purchasing preferences change. These numbers may be adjusted as necessary to test an alternate set of beliefs. The formula below represents this calculation. D n = S n – [(S n-1 + C' n-1 )-S n ] (1) where: D n – depletion rate; S n – September 1 hop stocks figures for year n; S n – 1 – September 1 hop stocks figures for year n – 1; C' n – 1 – total production of the US crop once processing loss has been accounted for. 2.2 Data Populations For this research, we analyzed USDA season average price (SAP), September 1 hop stocks and acreage for two data subpopulations. 1980-2000: The years 1980 to 2000 were marked by a lack of effective regulation over supply and the introduction of new hop varieties and products with greater efficiency. This was a time when four situations converged created a surplus market situation. 1) The third Federal Marketing Order had failed because as Folwell (1982) mentioned growers increased saleable quantities to take market share away from German growers (creating a permanent infrastructure production solution in the U.S. to a temporary German yield problem). 2) The new high-yielding alpha varieties, Nugget and Galena, were introduced to the market replacing lower-yielding alpha producing hops on a 1:1 basis increasing grower production efficiency of alpha-acid per acre. 3) Merchant proprietary products based on alpha-acid with greatly increase efficiencies were introduced to the market reducing demand by brewers. There were no effective forms of production or sales regulation during this period. Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ 57 4) Early proprietary alpha hop varieties were quietly introduced into the market under other variety names (i.e., red stripe Nugget). It is not possible to quantify the scale of this effect as the new acreage was reported as Nugget and there was no way to distinguish actual Nugget acreage from "red stripe" Nugget acreage (which many later speculated was one of the CTZ varieties). 1998-2020: This was a period during which branded proprietary varieties were introduced into the market. It was also a time during which the craft brewing industry reached a tipping point enabling its demands for proprietary aroma varieties for which they were willing to pay a premium to direct the industry. Since 1938, the U.S. hop industry had experimented with Federal Marketing Orders to regulate the saleable quantity of hops to brewers. All three Federal Marketing Orders that were enacted failed due to the lack of authority to effectively regulate grower activities. A fourth Federal Marketing Order was voted down in 2003 with concerns over the potential for the formation of a cartel expressed by one significant opponent (Federal Register 2020). Proprietary variety production, however, grew to such an extent that it effectively created the regulatory system the industry had long sought via the enforcement of intellectual property rights (IPR). Correlation however does not imply causation. The reverse, however, is not true. Causation implies some form of correlation (Goldthorpe 2001). 2.3 Bayesian Inference For each data set, we analyzed the effect of a hypothetical change in the independent variable upon the directional movement of the dependent variable (i.e. increase, decrease, no change) while keeping other variables stable (Hines 2015). We measured the effect of consecutive hypothetical identical changes to the independent variable upon the dependent variable to determine the probability of a change occurring because of the change. Limiting the possible numbers of outcomes increased the ability of our Bayesian inferential analysis to yield meaningful probability forecasts. A simple model can increase the chance for accuracy and reduce errors in forecasting by as much as 25% (Green and Armstrong 2015). The parameters of this analysis were limited to the directional movement of dependent and independent variables to limit false discovery data. More accurate probabilities resulted without compromising the integrity of the operation. Yekutieli (2012) concluded that while specifying a selection rule introduces an arbitrary element to Bayesian analysis, the selection rule was determined before the data were observed and carried out the same way as Bayesian inference. Our inferences were based on the Bayesian posteriors generated from a selection-adjusted analysis, which as Yekutieli (2012) noted led to a reduction in the forecasts possible from the available data (i.e. forecasts regarding directional price movements as opposed to forecasts regarding the probability of actual future prices). 58 Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ 2.4 The Bayesian Formula P(A|B) = (| ) () P(A) where: P(A|B) - probability of A occurring given that B was true; P(B|A) - probability of B occurring given that A was true; P(B) - probability of observing B; P(A) - probability of observing A; A and B - unique events. 3 RESULTS AND DISCUSSION The Bayesian inferential of analyses of the U.S. Season Average Price (SAP) resulted in the emergence of several sequences of data demonstrating that September 1 hop stocks and acreage did not follow market signals indicating reduced demand for hops. We discovered that when U.S. SAP decreased two years in a row during the period 1980-2000 it resulted in a 100% likelihood that September 1 hop stocks would increase, a trend that continued through the fourth consecutive year of U.S. SAP decreases (Table 1). Table 1: U.S. SAP decreased in consecutive years and the probability of an increase in U.S. September 1 hop stocks (1980-2000). DESCRIPTION Prior Probability Posterior Probability PPV Sensitivity Specificity U.S. SAP year n decreased; U.S. September 1 stocks increase 38,10% 41,67% 62,50% 41,67% 62,50% U.S. SAP year n decreased 2 consecutive years; U.S. September 1 stocks increase 41,67% 100% 61,54% 100% 0,00% U.S. SAP year n decreased 3 consecutive years; U.S. September 1 stocks increase 100% 100% 64,29% 100% 0,00% U.S. SAP year n decreased 4 consecutive years; U.S. September 1 stocks increase 100% 100% 66,67% 100% 0,00% Source: USDA NASS 2013, USDA NASS 2014, USDA NASS NHR 2000-2020 The data reflected that these two factors impacted market forces. There was only a slight reaction in the n year to a decrease in U.S. SAP. The prior probability of an increase in stocks of 38,10 percent changed to a posterior probability of 41,67 Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ 59 percent, a mild change. In the second consecutive year and beyond the posterior probabilities were 100 percent. The same was true of the sensitivity demonstrating the certainty of the true positive results. The PPV values introduced some room for doubt, but only marginally. The n year PPV value was 62,50 percent and increased by the fourth consecutive year to a mere 66,67 percent. Similar changes during the 1998-2020 period yielded very different results with very low probabilities of increases in September 1 hop stocks in years two, three and four of the calculations (Table 2). Following consecutive years of decreased U.S. SAP, the posterior probability (i.e., the likelihood) that stocks will increase resulting from these changes decreased to 21,69 percent by the fourth consecutive year of decreased U.S. SAP. The PPV value was high in the n year at 90,00 percent. This demonstrated a high degree of confidence in the posterior probability generated in that year. The PPV values decreased during the second, third and fourth consecutive year of decreasing U.S. SAP. By the fourth consecutive decrease in U.S. SAP, the PPV value was 60,00 percent. This is considered by our research to be a high value, but in conjunction with the direction of its movement provided by the other years measured in this research, we could better understand its relative significance. Table 2: U.S. SAP decreased in Consecutive Years and the Probability of an Increase in U.S. September 1 Hop Stocks (1998-2020). DESCRIPTION Prior Probability Posterior Probability PPV Sensitivity Specificity U.S. SAP year n decreased; U.S. September 1 stocks increase 43,48% 60,00% 90,00% 60,00% 85,71% U.S. SAP year n decreased 2 consecutive years; U.S. September 1 stocks increase 60,00% 44,68% 53,85% 87,50% 0,00% U.S. SAP year n decreased 3 consecutive years; U.S. September 1 stocks increase 44,68% 31,58% 57,14% 88,89% 0,00% U.S. SAP year n decreased 4 consecutive years; U.S. September 1 stocks increase 31,58% 21,69% 60,00% 90,00% 0,00% Source: USDA NASS 2013, USDA NASS 2014, USDA NASS NHR 2000-2020 60 Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ There exists a symbiotic relationship between acreage, production and stocks. Experiments regarding U.S. SAP data and its effect upon U.S. acreage yielded data that demonstrated that acreage lacked responsiveness to price signals indicating reduced demand during the 1980-2000 period (Table 3). There was a low probability of a decrease in acreage in response to consecutively decreasing prices. The year n posterior probability of a 37,50 percent likelihood of such a decrease increased to 50 percent by the fourth year of consecutive decreased U.S. SAP. Table 3: U.S. SAP decreased in consecutive years and the probability of U.S. acreage decreases (1980-2000). DESCRIPTION Prior Probability Posterior Probability PPV Sensitivity Specificity U.S. SAP year n decreased; U.S. acreage decrease 42,86% 37,50% 33,33% 37,50% 50,00% U.S. SAP year n decreased 2 consecutive years: U.S. acreage decrease 37,50% 36,55% 40,00% 44,44% 50,00% U.S. SAP year n decreased 3 consecutive years: U.S. acreage decrease 36,55% 38,59% 45,45% 50,00% 50,00% U.S. SAP year n decreased 4 consecutive years: U.S. acreage decrease 38,59% 42,99% 50,00% 54,55% 50,00% Source: USDA NASS 2013, USDA NASS 2014, USDA NASS NHR 2000-2020 The low PPV values of 33,33 percent in the n year increasing only to 50,00 percent in the fourth consecutive year of U.S. SAP decreases instilled confidence in the ability to forecast accurately using these results. These data are a result of the following events that occurred: (1) The newly introduced high yielding alpha varieties of Nugget and Galena created a Bertrand Supertrap reducing demand for U.S. hop acreage as efficiency was greatly increased on the farm. (2) The introduction of proprietary processed alpha- oriented products added to the effects of the Bertrand Supertrap. The hyper efficiency of these new products further reduced the demand for alpha hop products by brewers and the need for existing hop acreage. (3) The complete lack of effective regulation of production or saleable quantities, acreage and production meaning that laissez-faire economics were governing the market, which was in essence a free for all. (4) The homogeneity of hop products marketed between 1980 and 2000. Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ 61 Even some branded proprietary varieties that existed at the time (i.e., a proprietary variety referred to as Red Stripe Nuggets so it could be sold as a Nugget when it was not) found their way into the homogenous product stream offering a disproportionate advantage to the morally impaired growers who did not shy away from selling an intentionally mislabeled product. The data from the 1998-2020 period demonstrated that the opposite was true during that time, which indicated a greater likelihood of responsiveness to market price signals (Table 4). The data during the period between 1998 and 2020 suggested a high responsiveness to consecutive decreased U.S. SAP. Such strong data were not evidenced during the other data subpopulations measured with respect to U.S. SAP and U.S. acreage. In year n, in response to a single decrease of U.S. SAP the probability of acreage decreasing increased from the prior probability of 43,48 percent to a posterior probability of 71,43 percent. After another year of decreased U.S. SAP, the posterior probability representing the likelihood of an acreage decrease jumped to 89,86 percent. In the third and fourth consecutive years the likelihood soared yet again to 97,11 percent and 99,26 percent probability. These data represented not only a high degree of correlation but causation between the two data points under these circumstances. Table 4: U.S. SAP Decreased in Consecutive Years and the Probability of U.S. Acreage Decreases (1998-2020). DESCRIPTION Prior Probability Posterior Probability PPV Sensitivity Specificity U.S. SAP year n decreased; U.S. acreage decrease 43,48% 71,43% 50,00% 71,43% 68,75% U.S. SAP year n decreased 2 consecutive years: U.S. acreage decrease 71,43% 89,86% 54,55% 75,00% 68,75% U.S. SAP year n decreased 3 consecutive years: U.S. acreage decrease 89,86% 97,11% 58,33% 77,78% 68,75% U.S. SAP year n decreased 4 consecutive years: U.S. acreage decrease 97,11% 99,26% 61,54% 80,00% 68,75% Source: USDA NASS 2013, USDA NASS 2014, USDA NASS NHR 2000-2020 62 Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ The PPV remains strong with values going from 50,00 percent in the n year to 61,54 percent in the fourth consecutive year introducing reasonable doubt into the equation in what otherwise appeared to be near certainty with regards to acreage reduction. The high degree of sensitivity in these data with values ranging from 71,43 percent in the n year and increasing to 80,00 percent in the fourth consecutive year was of interest in that it suggested a high True Positive (TP) rate. These results were significant. They represented the first occurrence of a situation in which we can see that the DSR effect was not present indicating that something has changed with regards to how the market operates. There was an increased degree of control making acreage more responsive to market forces. 4 CONCLUSIONS The period 1980–2000 contained data that represented surging stock levels and what we have referred to as the DSR. Stock levels were among the underlying market characteristics that affected price volatility. Data suggest the disequilibrium created by the DSR has been responsible for market imbalances and the boom-and- -bust cycles recorded in the hop industry. The increase in the concentration of power within the hop industry changes market dynamics. SAP data strong merchant oligopoly during times of deficit to a strong brewer oligopsony during times of surplus combined with forced contracting that is later renegotiated respectively contributes to the DSR. Reduced contract revenue following renegotiation and/or reneging resulting from decreased prices enables lower profits than would otherwise be possible on the spot market without the presence of contracts (Cabral and Villas-Boas 2005). The period 1998–2020 produced data that demonstrated the absence of the DSR during that time representing a significant change in market dynamics. One of the primary differences between the 1980–2000 and 1998–2020 periods was the ability to effectively regulate the production and saleable quantity of proprietary hop varieties using IPR. Intellectual property (IP)-related constraints to the market affected planting decisions (Stiglitz and Dixit 1977). Patented varieties represent the potential for legal monopoly control over a product. IPR empower individuals to regulate acreage, production, and therefore stocks of the varieties they own. The DSR represented the natural dynamics of the hop market for centuries. The high concentration of proprietary hop varieties in the US industry altered those dynamics. It is still unknown whether their presence has completely removed the DSR or created an extended DSR that will reveal itself over a longer period. The authors speculate the latter will be the case, but further research in the future will be necessary to confirm this. Hmeljarski bilten / Hop Bulletin 30 (2023) ______________ 63 5 REFERENCES Bayarri, M.J., Berger, J.O. (2004): The Interplay of Bayesian and Frequentist Analysis. Statistical Science, 19, 58-80. DHWV (2012): Deutscher Hopfenwirtschaftsverband e.V. Market Report. International Hop Growers’ Convention. Available at http://www.hmelj-giz.si/ihgc/act.htm (accessed May 11, 2022). Folwell R.J. (1982): The U.S. Hop Marketing Order: The Price of Success is Misunderstanding. Exhibit 28 USDA Federal Hop Marketing Order Hearings. Folwell R.J., Mittelhammer R.C., Hoff F.L., Hennessy P.K. (1985): The Federal Hop Marketing Order and Volume-Control Behavior. Agricultural Economics Research 37: 17-32. Goldthorpe J. (2001): Causation, statistics and sociology. European Sociological Review, 17: 1–20. Green K., Armstrong J. (2015): Simple versus complex forecasting: The evidence. Journal of Business Research, 68: 1678–1685. Henning J.A., Gent D.H., Twomey M.C., Townsend M.S., Pitra N.J., Matthews P.D. (2015): Precision QTL mapping of downy mildew resistance in hop (Humulus lupulus L.). Euphytica, 202: 487-498. Hines K. (2015): A primer on Bayesian inference for biophysical systems. Biophysical Journal, 108: 2103–2113. Stiglitz J.E., Dixit A.K. (1977): Monopolistic competition and optimum product diversity. The American Economic Review, 67: 297–308. MacKinnon D., Pavlovič M. (2021): Hoponomics: Market analysis applied to the hop industry 1980-2020. Hamburg: Verlag Dr. Kovač, 197 p. MacKinnon D., Pavlovič M. (2022): The delayed surplus response for hops related to market dynamics. Agricultural economics, 68, 8: 293-298. Pavlovič M., MacKinnon D., Naglič B. (2022). Study of market supply statistics in trading of hops. Hmeljarski bilten, 29: 85-96. USDA NASS. (2013): WA Hops 1915-2013. Available at www.nass.usda.gov- Statistics_by_State-Washington- Publications-Historic_Data-hops-wahops.pdf (accessed April 28, 2022). USDA NASS. (2014): Hop Stocks Held by Growers, Dealers, and Brewers, United States (1,000 Pounds). Available at www.nass.usda.gov-Statistics_by_State-Washington- Publications-Historic_Data-hops-ushopstk.pdf (accessed April 28, 2022). USDA NASS. (2000-2020): NHR. Available at www.nass.usda.gov-Statistics_by_State- Washington-Publications-Hops-index.php (accessed March 29, 2022). Yekutieli D. (2012): Adjusted Bayesian inference for selected parameters. Journal of the Royal Statistical Society. Vol. 74, No. 3: 515-541.