Changing market dynamics and value-added premiums in southeastern feeder cattle markets

ABSTRACT

The southeast region from Kentucky and Virginia south to Florida is home to more than 5 million beef cows, representing a significant source of farm income. Because very little feedlot capacity exists in the region, feeder cattle sales represent the majority of beef cattle receipts. This work examined numerous factors that affect feeder cattle prices including the effect of changes in com price, the dynamics of BW uncertainty in video auctions, and price premiums associated with age and source verification and cattle selling certified natural. Sale data from Internet sales and preconditioned feeder cattle sales were made available by a large cattle marketing firm in the region from 2008 to 2011. A hedonic model was used to examine the effect of cattle characteristics and market factors on the selling price of feeder cattle. Evidence was found that corn price negatively affected feeder cattle prices, but the magnitude was found to be smaller than in previous research. Premiums for age and source verification were found to be moderate at $15.00 per animal, which was largely consistent with estimates from other regions of the United States. Premiums for cattle selling certified natural were estimated to be around $17.00 per animal. Finally, in contrast to previous research, results suggest that market incentives with respect to existing price-BW relationships have changed such that incentives to underestimate BW in video auctions may not have existed during this time period.

Key words: corn-price effect, feeder cattle, livestock auction, price premium, price slide

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

Numerous factors have been established to affect feeder cattle prices including expected fed cattle price, corn price, BW, lot size, and many factors related to feeder cattle health, quality, and market conditions. Because corn is the major input for feedlots, corn price has long been established to be inversely related to feeder cattle price. However, the last several years have been associated with substantially higher corn prices and increased price variability. It is possible that the dynamics of the relationship between corn and feeder cattle prices have changed in the current corn-market environment.

One effect of higher corn prices tends to be a narrowing of feeder cattle price differences by BW. In other words, there is less of a price decrease as feeder cattle become heavier. This occurs during high-feed-price time periods because feedlots look to place heavier cattle during this time. At the same time, feeder cattle are often offered for sale via video where BW is not known with certainty. In those cases, cattle are typically advertised with a base BW, and an artificial price slide is offered to adjust the price downward when actual BW exceeds the expected BW. It is not clear that these predetermined price slides have evolved with the existing priceBW relationships in the marketplace.

Numerous value-added characteristics have also affected feeder cattle prices over the last several years. Price premiums for feeder cattle selling age and source verified have been estimated by research, but these estimates have not been consistent and have not been specific to the Southeast (Kellom et al., 2008; Schumacher et al., 2012; Williams et al., 2012; Zimmerman et al., 2012). Furthermore, much less research has examined the price premium for cattle selling certified natural, which ap- pears to be a growing market in the United States.

Given these considerations, there were 3 primary objectives of this research. The first was to examine the dynamics of feeder cattle price and corn price relationships in the current corn market environment. The second was to evaluate the price premiums paid for age and source verification and certified natural in Internet auctions in the Southeast. The third objective was to examine current price differentials by BW and price slides to determine what marketing incentives exist in the current market.

MATERIALS AND METHODS

A strength of this research is the unique pair of data sets obtained by the authors. A large cattle marketing firm in Kentucky made available data from their Internet sales and preconditioned feeder cattle sales from January 2008 to April 2011. The data set from the Internet sales included more than 1,600 lots of feeder cattle from Alabama, Florida, Georgia, Indiana, Kentucky, North Carolina, Ohio, Tennessee, Virginia, and West Virginia. Electronic files included price, BW, and lot size information and were supplemented by manual entry of data from sale catalogs that were made available to buyers on the sale day and included BW, cattle description, weigh and sale conditions, implants, location, and any further marketing claims.

The second data set contained more than 1,300 lots from preconditioned feeder cattle sales. Kentucky Certified Pre-conditioned for Health (CPH) sales are held 4 to 7 times annually for producers to sell cattle managed under a uniform health program. General requirements include that calves be weaned a minimum of 45 d before sale, be bunk and trough broke, castrated, and dehorned and healed. In terms of the health program, calves were to be vaccinated for bovine respiratory virus, parainfluenca-3, bovine viral diarrhea, and infectious bovine rhinotracheitis of which the booster was required to be a modified live product. Mannheimia haemolytica vaccine is also required. In addition, there is a monetary guarantee that no heifers are bred and no males are intact. Calves from multiple producers are sorted by sex, BW, frame, and color and sold in commingled groups. Price, sale BW, lot size, sex, and color sort were all included in the CPH data set.

Both data sets were augmented with additional data from several sources. Historical diesel-fuel price data were available from the EIA (2011). According to EIA, their data are collected each Monday through a phone survey. It was assumed that the Midwest diesel price from the EIA (2011) survey applied to any cattle sales occurring during that same week.

Daily corn and live cattle futures prices were available from the Livestock Marketing Information Center (LMIC, 2011), which databases futures prices from the CME Group Inc. In the case of both the Internet and CPH sales, the closing price for the nearby corn futures contract on the day of the sale was used. Because feeder cattle are likely to be on feed for several months, a determination had to be made as to which deferred live cattle contract was relevant for a given lot of feeder cattle. The Livestock Marketing Information Center tracks data from Kansas State University's monthly Focus on Feedlots Survey (Waggoner, 2011). During the research period, average steer slaughter BW was 606 kg, whereas average heifer BW at slaughter was 552 kg (Waggoner, 2011). These average slaughter BW were used to estimate slaughter dates, assuming the ADG of steers was 1.59 kg and heifers was 1.43 kg. Given an estimated slaughter date, the price of the next expiring live cattle futures contract was used.

Hedonic regression models were used to estimate the determinants of feeder cattle prices in the Internet sales and in the CPH sales. Hedonic models are common in the literature as a way to estimate the price effects of various feeder cattle characteristics (Bulut and Lawrence, 2007; Kellom et al., 2008; Schulz et al., 2010; Williams et al., 2012; Zimmerman et al., 2012). The dependent variable in the Internet sales regression was bid price, defined as the price the cattle actually sold for on the day of the Internet sale, not including any price slide adjustments. This was calculated using the final price, actual BW, and advertised slide. In the CPH regression, the dependent variable was sale price, simply the price the cattle sold for on the day of the CPH sale.

The models below were estimated using the "proc model" procedure in SAS (SAS Institute Inc., Cary, NC):

... (1)

where all variables are specified as described in Table 1. Umonth is a series of binomial variables for each month excluding January, ^location is a series of binomial variables for each state in which cattle originated except Tennessee. V12cattle type is a series of binomial variables for each cattle type except black or black-white face cattle. Uptime is a continuous variable to capture time trend; the first sale date is 1 and a sale 30 d later would have a time value of 30, and so on.

...(2)

where all variables are specified as described in Table 1. V^month is a series of binomial variables for each month excluding January, and Vgcattle sort is a series of binomial variables for each CPH cattle sort group except black cattle. Retime is a continuous variable to capture time trend; the first sale date is 1 and a sale 30 d later would have a time value of 30, and so on. Table 1 provides a description of each variable used in both models.

A Durbin-Watson test for auto-correlation yielded a t-statistic outside the accepted range, and a regression of the squared residuals from the models against dependent and independent variables suggested the presence of heteroskedasticity. To correct for heteroskedasticity and autocorrelation, a Newey-West estimation was employed. Models were also estimated using basic Ordinary Least Squares and the Robust Estimator in SAS (SAS Institute Inc.) with very little effect on results.

Models were tested for multi-collinearity through a variance of inflation (VIF) test. Results suggested that VIF statistics for corn and live cattle price in both models were slightly higher than desired. However, because the problem suggested was small and both were crucial variables explaining feeder cattle price, excluding one of them seemed inappropriate and both were retained in the models. The only case where VIF statistics suggested major concern was in initial model specifications where both slide 1 and slide2 were included. For clarity, slidel is the price slide for the first 22.7 kg over the base BW and slide2 is the price slide once the payweight exceeds 22.7 kg over the base BW. Furthermore, VIF statistics were so high (greater than 200) that not addressing this problem did not seem to be an option. The only logical solution to the problem was to exclude the second slide variable, slide2, from the model. Once deleted, VIF statistics returned to highly acceptable levels.

RESULTS AND DISCUSSION

Tables 2 through 5 contain descriptive statistics for the 2 data sets. As can be seen in Table 2, cattle marketed in Internet sales were largely sold in load lot quantities. They were also relatively heavy with an average BW of 362 kg. During the time period examined, corn price ranged from $3.06 to a high of $7.55, but averaged $4.56/25.4 kg (bushel). As can be seen in Table 3, 87.3% of the lots were steers, which partially explains the heavy average BW at marketing. Cattle were sold in all months of the year and came from 10 states. The majority of the cattle were advertised as black and black-white faced. In terms of value-added markets, 7.3% were age and source verified only, 1% were natural only, and 2.8% were both age and source verified and natural. Natural cattle were free of implants and ionophores and had not received antibiotics.

As expected, cattle selling in preconditioned CPH sales were lighter, with an average BW of 279 kg. As can be seen in Table 4, lot size was considerably smaller, averaging around 19, but ranging from 1 to 286 cattle. Average corn price for the CPH data was $4.54/25.4 kg with a range of $3.18 to $7.54. Table 5 depicts a relatively even split between steers and heifers. Cattle were sold during the months of January, February, March, April, June, November, and December, which is simply of function of when CPH sales are offered. Color and breed sorts were variable, but the most common sorts were black, Charoláis cross, smoke, and black cross.

Results from the Newey-West estimation are reported in Tables 6 and 7. General explanatory power of both models was quite strong with coefficients of determination more than 90% for the Internet sale price model (Equation 1) and more than 77% for the CPH sale model (Equation 2). General findings were logical and consistent with the literature in terms of signs and significance of key variables from similar studies (Kellom et al., 2008; Schulz et al., 2010; Williams et al., 2012; Zimmerman et al., 2012).

Although results from this research are largely consistent with theory and previous literature, it is important that one recognize the data used were regional. Data from the Internet sales included cattle from 10 states, primarily in the Southeast. Data from CPH sales were from a single location in Kentucky. Although the authors feel the quality of the data is very good, potential for selection bias certainly exists. For this reason, results should not be generalized as representative of the entire US feeder cattle market.

Most research in this area has been conducted during time periods when corn prices were considerably lower and less volatile. The time period examined in this research provided the opportunity to examine feeder cattle prices over a very wide range of corn price levels. Current findings suggest that feeder cattle prices were negatively affected by corn price as they have been demonstrated to be by others (Buceóla, 1980; Trapp and Eilrich, 1991; Dhuyvetter and Schroeder, 2000; Burdine et al., 2004). A $1 increase in corn price was associated with a $2.97 decrease in feeder cattle price per 45.36 kg in Internet sales (Table 6). With a corn price standard deviation of $1.17, a one standard deviation increase in corn price was associated with a decrease in price of $3.47/45.6 kg. In the CPH sale data set where BW was lower, a one unit change in corn price was associated with a $4.02 change in price per 45.36 kg (Table 7). With a corn price standard deviation of $1.29, a one standard deviation increase in corn price was found to be associated with a $5.17 decrease in price per 45.6 kg. The effect of corn price has generally been found to be smaller for heavier feeder cattle (Buceóla, 1980; Trapp and Eilrich, 1991; Dhuyvetter and Schroeder, 2000; Burdine et ah, 2004), a finding that was supported by this research as corn price effects were found to be larger in CPH sales, where average BW was lower.

Parameter estimates on corn price in this research are slightly lower than those found in previous literature (Buceóla, 1980; Trapp and Eilrich, 1991; Dhuyvetter and Schroeder, 2000; Burdine et ah, 2004). However, this difference between our research and literature is likely explained by the uniqueness of this data set and implications of feeder cattle markets absorbing drastically higher corn prices. Because corn-price levels were higher during the time period of this research (2008 to 2011), greater incentives existed for feedlots and backgrounders to explore alternative feeds and feeding systems as was discussed in Anderson and Trapp (2000). As we consider the implications for the time period of the current research, changes in corn-market dynamics from increased ethanol production have no doubt changed since 2000.

One of the unique aspects of Internet sales is the element of uncertainty that is present. The cattle are not seen live, BW is not known with certainty, and other factors are largely only known to the extent that they are visible via video or revealed by the consignor. This allows for analysis of unique factors, most interestingly the artificial price slide that is used to adjust price downward for cattle that weigh above their advertised base BW in the sale catalog. A clear strength of this data set was the ability to examine data ex-ante, meaning it provided the opportunity to examine actual bid prices, rather than final slide-adjusted prices alone. This provided the opportunity to examine the effect of the price slide on sale price and compare this to the actual difference in price for cattle of different BW during the research period.

Price slides have been used for years to cope with BW uncertainties and remain the tool of choice today. However, producers must remain aware of the incentives that exist in the marketplace. The artificial price slides that are offered in sale catalogs are intended to protect buyers from paying a higher price for cattle because they were advertised at a lower BW. The price slide essentially provides a price penalty for cattle that weigh more than they were advertised. Brorsen et al. (2001) found that price slides were typically not large enough to provide a disincentive to underestimate BW. Hence, they found evidence that cattle typically weighed more than advertised in video-based sales. However, little evidence of BW underestimation was present in the data set employed in this analysis. In the Internet sales examined in this research, average BW were only 0.81 kg greater than the advertised base BW, less than those observed by Brorsen et al. (2001). Although numerous factors could explain this difference, changing market incentives are likely to have been at play.

The absolute smallest price slide offered in the Internet data set was $4.00/45.4 kg, and this was generally the price slide offered for cattle that weighed near the 363-kg average. For an incentive to exist for producers to underestimate BW, the actual negative price effect of BW in the market would need to exceed $4.00/45.4 kg. Although this incentive may have existed during 2001 when corn prices were lower and the Brorsen et al. (2001) study was conducted, our data suggest that it did not exist from 2008 to 2011. Note the parameter estimate on the BW variable in Table 6; a 45.4-kg increase in BW was associated with a price decrease of only $2.55/45.4 kg. Cattle that were sold on an artificial price slide of $4.00/45.4 kg and weighed 45.4 kg more than their base BW would receive a price discount of $4.00. However, results suggest that same group of cattle would have only sold for $2.55 less during this research period.

Take for example a group of steers that were advertised with a base BW of 363 kg, an artificial price slide of $4.00/45.4 kg, and sold for $100.00/45.4 kg. Now suppose that they actually weighed 408 kg. With the artificial price slide, the final selling price for the cattle would be adjusted downward by $4.00 to $96.00/45.6 kg and the steers would sell for $864 each. However, if the cattle had been advertised as weighing 408 kg from the start, the actual market price slide was only $2.55. This would have resulted in an expected selling price of $97.45/45.4 kg and a value per animal of $877.05. Put simply, these cattle would have brought more if advertised at the higher BW from the beginning.

This scenario suggests that the market incentive to underestimate BW found in Brorsen et al. (2001) did not exist during the time period of this research. The economic environment during the data collection period actually provided much greater incentive for producers to know the BW of their cattle because the typical price slide offered in the catalogs provided a greater price penalty than the market itself.

These findings suggest that price slides, as they are used to adjust prices for BW differences, have not evolved with the actual price-BW relationships in the marketplace. Brorsen et al. (2001) noted that price slides amounted to an "option" for sellers. When price slides are less steep than actual price-BW relationships in the marketplace, consignors have incentive to deliver heavier cattle; Brorsen et al. (2001) found that tendency. It would appear that this incentive has changed over the last several years and is evidenced by a much smaller tendency to underestimate BW in this work.

In a sense, the market is more efficient than it used to be because sellers do not have a strong market incentive to underestimate BW. However, because of the flexible nature of the delivery times, the current system may offer incentive for buyers to delay delivery. Examination of sale catalogs from 2008 through 2011 suggests that most delivery ranges are about 1 wk, but 2 or 3 wk were offered in some cases. In cases where a great deal of flexibility is offered, it would not be surprising to see delivery dates pushed back and cattle BW start to increase because this would actually benefit the buyers. Of course this incentive is probably less of a problem than the one that Brorsen et al. (2001) discovered because consignors can simply tighten up delivery windows. But to do this, they must be aware of why doing so makes economic sense.

Finally, as the price effects of age and source verification were the initial motivation for this work and the primary interest of stakeholders, further discussion of these estimates is warranted. Results in Table 6 provide evidence that price premiums existed in Internet sales, because a significant positive price relationship was found to exist of $1.83/45.4 kg for age and source verification, almost $15.00 per animal. This result is very consistent with previous estimates in the literature (Kellom et al., 2008; Zimmerman et al., 2012) but inconsistent with Williams et al. (2012), where no significant premium was found, and considerably lower than the estimates made by Schumacher et al. (2012). This research provided the first estimate of age and source verification for cattle in the Southeast, and it is noteworthy that estimates were consistent with 2 prior studies using data sets from other regions of the United States (Kellom et ah, 2008; Zimmerman et ah, 2012).

Natural calves, those sold without implants, ionophores, or antibiotics, were associated with a significant price premium of $2.16/45.4 kg, and cattle selling as both age and source verified and certified natural were associated with a price premium of $3.99/45.4 kg. These were considerably higher premium levels than those found in the work of Zimmerman et al. (2012).

Although the price premiums for cattle selling as natural were larger than those associated with age and source verification, this may actually be less appealing to producers. The primary cost associated with age and source verification is time spent maintaining records. In the case of selling cattle as certified natural, the primary cost is likely increased production costs due to lost performance. Without the aid of implants, rates of gain and BW sold are likely to be lower, resulting in lower revenues. As producers consider the cost-benefit of selling natural calves, these data suggest that the benefit may only be around $17.00 per animal on a 363-kg steer. This benefit must be carefully weighed against the cost of additional feed needed or fewer kilograms marketed as a result of lower feed conversion. Cattle selling that were both age and source verified and certified natural saw price premiums near $32.00 per animal. Increased understanding of price premiums in value-added markets provides valuable information to both producers and marketing professionals.

IMPLICATIONS

The long-held negative relationship between corn price and feeder cattle price held during a time period associated with considerably higher and more volatile corn prices, with some evidence that the magnitude has decreased. Second, changing market conditions have eliminated the incentive for consignors to underestimate BW when cattle are sold on traditional prices slides and likely increased the value of consignors knowing the BW of the cattle being sold. Finally, the price premium for age and source verified cattle in the Southeast was observed to be consistent with prior work but found to be considerably higher for cattle selling in Internet sales that were also certified natural.

LITERATURE CITED

Anderson, J. D., and J. N. Trapp. 2000. The dynamics of feeder cattle market responses to corn price change. J. Agrie. Appl. Econ. 32:493 505.

Brorsen, B. W., N. Coulibaly, F. G. C. Richter, and D. Bailey. 2001. Feeder cattle price slides. J. Agrie. Resour. Econ. 26:291-308.

Buceóla, S. T. 1980. An approach to the analysis of feeder cattle price differentials. Am. J. Agrie. Econ. 62:574-580.

Bulut, H., and J. D. Lawrence. 2007. The value of third-party certification of preconditioning claims at Iowa feeder cattle auctions. J. Agrie. Appl. Econ. 39:625-640.

Burdine, K. H., A. L. Meyer, and L. J. Maynard. 2004. Understanding the market for Holstein steers. Staff paper no. 447. Dept. Agrie. Econ., Univ. Kentucky, Lexington.

Dhuyvetter, K. C., and T. C. Schroeder. 2000. Price-weight relationships for feeder cattle. Can. J. Agrie. Econ. 48:299-310.

EIA. 2011. Regional diesel fuel prices. Energy Info. Admin., US Dept. Energy. Accessed Nov. 29, 2013. http://www.eia.gov/oog/info/ wohdp/diesel. asp.

Kellom, A., J. Patterson, J. Vanek, M. Watts, and M. Harbac. 2008. The effects of age and source verification of calves on value received on Superior Livestock Video Auctions. Proc. West. Sect. Am. Soc. Anim. Sei. 59:137-139.

LMIC. 2011. Futures prices databased from Chicago Mercantile Exchange. Livestock Mark. Info. Center. Accessed Nov. 29, 2013. www.lmic.info.

Schulz, L., K. C. Dhuyvetter, K. Harborth, and J. Waggoner. 2010. Factors Affecting Feeder Cattle Prices in Kansas and Missouri. Dept. Agrie. Econ., Kansas State Univ., Manhattan.

Schumacher, T., T. C. Schoeder, and G. T. Tonsor. 2012. Willingness-to-pay for calf heath programs and certification agents. J. Agrie. Appl. Econ. 42:191-202.

Trapp, J., and F. Eilrich. 1991. An analysis of factors affecting Oklahoma City feeder cattle basis. Pages 180-192 in Applied Commodity Price Analysis, Forecasting, and Market Risk Management. NCR-134 Conf. Comm., Chicago, IL.

Waggoner, J. 2011. Focus on feedlots. Dept. Anim. Sei. Ind., Kansas State Univ. Accessed Nov. 29, 2013. http://www.asi.k-state.edu/ about / newsletters/focus-on-feedlots/.

Williams, G. S., K. C. Râper, E. A. DeVuyst, D. Peel, and D. McKinney. 2012. Determinants of price differentials in Oklahoma value-added feeder cattle auctions. J. Agrie. Resour. Econ. 37:114-127.

Zimmerman, L. C., T. C. Schroeder, K. C. Dhuyvetter, K. C. Olson, G. L. Stokka, J. T. Seeger, and D. M. Grotelueschen. 2012. The effect of value-added management on calf prices at Superior Livestock Auction video markets. J. Agrie. Resour. Econ. 37:128143.

K. H. Burdine,*1 L. J. Maynard,* G. S. Halich,* and J. Lehmkuhlert

1 Department of Agricultural Economics, and fDepartment of Animal and Food Sciences,

University of Kentucky, Lexington 40546

1 Corresponding author: [email protected]