EQUINE

Association between the purchase price of Thoroughbred yearlings and their performance during the 2- and 3-year-old racing seasons
EQUINE
avj_829 388..393

MA Jackson,* AL Vizard, GA Anderson, AF Clarke and RC Whitton

Objectives Describe the association between the purchase price of Thoroughbred yearlings sold in Australia and racing performance as 2- and 3-year-olds. Methods Race performance data of 2773 Thoroughbred yearlings sold at auction during 2003 were collected. Associations between purchase price and the probability of starting, the number of race starts and the prize money earned were examined. Results In total, 2206 (79.6%) horses started a race. The mean number of race starts was six and the mean prize money earned was A$24,420. A total of 1711 (61.5%) horses earned prize money, 402 (14.4%) earned more than their purchase price, 312 (11.2%) earned more than A$40,000, the estimated cost of training, and 142 (5.1%) earned A$40,000 more than their purchase price. There was a positive association between purchase price category and the probability of starting, number of starts, earning prize money and earning greater than A$40,000 (P < 0.001). Purchase price category was negatively associated with the probability of earning greater than the purchase price (P < 0.001). The proportion of horses earning greater than the purchase price plus $40,000 was signicantly dierent (P = 0.03) among the ve price categories. Conclusion Yearling purchase price was positively associated with all race performance outcomes measured and researchers examining the race performance of yearlings purchased at sales should consider including purchase price when modelling. The Thoroughbred yearling market in Australia behaves in a similar manner to the United States market; owners pay a premium to enter the sport of racing and an additional premium in the quest to own a champion. Keywords economics; horse racing; racing career; Thoroughbreds; yearlings
Aust Vet J 2011;89:388393 doi: 10.1111/j.1751-0813.2011.00829.x

cally identied rst phalanx fragments, history of arthroscopy, sires stud fee, racing performance of the dam and racing performance of siblings have been consistently shown to aect the value of a yearling at sale.4,5,7,11,12 An economically rational investor would be prepared to pay no more for a yearling than the net present value of the expected life-time earnings of the horse plus the net present value of the residual value of the horse once its racing career has nished.13 Studies of the sale and subsequent performance of Thoroughbred yearlings in the United States have concluded that this market does not behave in a economically rational manner and that nancial returns from owning racehorses are negative in aggregate.13,14 Heckerman15 found that the average prize-money earnings of Thoroughbred racehorses in the United States increased with increasing purchase price. Heckerman15 also showed that the proportion of horses earning more than their purchase price decreased with increasing price. Only 6% of horses purchased for US$100,000 or more managed to earn more than their purchase price during their racing career, much lower than the 49% of horses purchased for US$10,000 or less.15 Data on purchase price and subsequent performance outside of the United States market are scant and it may be that other Thoroughbred yearling markets do not behave in the same manner as the North American market. In an Australian study to prole the racing performance of 1804 Thoroughbreds, More reported that 39.5% earned no prize money and 86.9% did not earn enough to cover estimated training costs in the rst year of racing,16 but these results were not linked to the purchase price of the horse. In a United Kingdom study of 1022 Thoroughbreds, 5% of 2-year-old and 17% of 3-year-old horses earned enough prize money to cover the costs associated with training,17 but again a relationship with purchase price was not examined. Recent studies have examined the association in yearlings between performance and orthopaedic ndings identied radiographically.1820 These studies have found no or few signicant associations between abnormalities observed on radiographs and subsequent race performance. Methods to improve the ability to identify true associations include increasing the number of yearlings examined and accounting for confounding variables. The largest study to date included purchase price in the statistical model and demonstrated an association between four radiographic ndings and decreased race performance.18 Other studies have found fewer signicant associations and involved lower numbers of yearlings and did not account for purchase price.19,20 Yearling purchase price data are readily available but there are limited data on its association with performance.1315 In order to determine whether the Australian yearling market behaves in a similar manner to the United States market and if yearling sale price needs to be considered in studies using future race performance as an outcome, we examined the association between the purchase
2011 University of Melbourne Australian Veterinary Journal Australian Veterinary Association

horoughbred breeding and racing is a large industry1,2 of which the sale of yearling horses is a core component. For example, during the 200203 season, 4057 yearlings were sold at auction in Australia for a total of approximately A$167 million.3 A number of studies have investigated the factors that inuence yearling purchase price.412 Macroeconomic factors such as available prize money, international buyers, incentive schemes and tax benets have been shown to have an eect on the purchase price of yearlings.5,8,9,12 Yearling-specic variables such as age or month foaled, radiographi*Corresponding author: Melissa.Jackson@nmit.vic.edu.au Faculty of Veterinary Science, University of Melbourne, Werribee, Victoria, Australia

388

Australian Veterinary Journal Volume 89, No 10, October 2011

EQUINE
price of 2773 Thoroughbred yearlings sold in Australia and their subsequent racing performance as 2- and 3-year-olds. Materials and methods Study population There are two major bloodstock companies in Australia: Magic Millions Pty Ltd and William Inglis and Son Ltd. They held 11 auction sales in Australia in 2003 specically for the sale of Thoroughbred yearlings.Yearlings in this study were sold at one of eight auctions held by either of these bloodstock companies and accounted for 68% of yearlings sold and 92% of the gross sale value of yearlings in Australia in that year.3 Yearlings that were entered for sale and then consequently withdrawn from the auction or those that were passed-in at sale were excluded from the analyses. Sale catalogues, provided by the bloodstock companies, were used to obtain identifying information for each horse such as lot number, sex, colour, foaling date, sire, dam and grand sire. This information was then cross-matched with the Australian21 and New Zealand22 stud books to determine if a horse had been registered for racing purposes or exported from Australia. Purchase price categories The purchase price of each yearling was acquired from the relevant bloodstock company website23,24 and allocated to one of ve categories: (1) $010,000; (2) $10,00120,000; (3) $20,00150,000; (4) $50,001 100,000; and (5) >$100,000. These groups approximated the quintiles for purchase price and represent the likely potential spending categories of purchasers. Race performance For all horses in the study, 2- and 3-year-old racing performance records from 1 August 2003 to 31 July 2005 were sought. Racing information that was recorded included: gelding reported, starting in a race, number of race starts, number of race places (combined number of rst, second and third places attained), prize money earned and change of trainer. Race performance information for horses racing in Australia was retrieved from the database of Racing Information Services Australia.25 For horses that were exported, racing performance records were retrieved from the relevant racing association through personal communication or via their website.2629 All overseas prize money earned was converted into Australian dollars30 and all money is reported in Australian dollars. In addition to the race performance outcomes listed above, the following binary performance outcomes were created for analysis: starting in at least one race, earning any prize money, earning prize money greater than the estimated cost of keep, earning prize money greater than the purchase price, and earning prize money greater than the combined estimated cost of keep and purchase price. Estimated costs The cost of keeping a racehorse in Australia during the rst full 2 years of racing was estimated at $40,000 by averaging costs per month quoted by companies syndicating horses of similar value to the mean purchase price of the study population group.3133
2011 University of Melbourne Australian Veterinary Journal Australian Veterinary Association

Statistical analysis Logistic regression and the likelihood ratio test were performed to assess the interaction between sex (male and female) and purchase price category for the various binary outcomes. The interaction between sex and purchase price category for prize money earned (log(dollars+1)) was assessed with linear regression. For the outcomes of number of race starts and number of race places, the interaction between sex and purchase price category was assessed by the use of the likelihood ratio test derived from negative binomial regression.34 Sex was also assessed as a potential confounder of the association between outcome and purchase price category by comparing the four coecients corresponding to the purchase price categories in a baseline model that included only purchase price category with those in a model that also included sex. Confounding was deemed to be present if a coecient changed by more than 20%.35 Sex and purchase price category were coded as indicator variables. Associations among the proportion of horses in each purchase price category being gelded post-sale, starting in at least one race, earning any prize money, earning prize money greater than the estimated cost of keep, earning prize money greater than the purchase price and earning prize money greater than the combined estimated cost of keep and purchase price were determined using Fishers exact test (SPSS version 16.0, SPSS Inc. and Stata version 9.2, StataCorp). The relationship between the number of race starts, the number of race places and the prize money earned for each of the ve purchase price categories was determined using a Kruskal-Wallis test (SPSS version 16.0). All race performance results are for the combined 2- and 3-year-old race seasons. A two-sided P-value <0.05 was considered statistically signicant. Results Results are reported as mean (median; range) unless otherwise stated. Horse signalment There were 1641 (59.2%) colts, 1130 (40.8%) llies and 2 (0.1%) geldings sold at the eight yearling sales (Table 1). The mean age at the time of sale was 1.4 years (1.4; 1.11.8). By the end of their third year, 1149 (70.0%) of the colts were reported as having been gelded post sale. The proportion of geldings in each of the purchase price categories, 15, respectively, was 75.8% (200/264), 73.4% (229/312), 72.3% (404/559), 68.6% (216/315) and 52.4% (100/191). As purchase price category increased there was a decrease in the number of colts reported as being gelded (P < 0.001). After removing category 5 from the analysis there was no dierence in the proportion of male horses gelded in categories 14 (P = 0.27). Sale results The mean purchase price for the 2773 horses was $55,299 ($30,000; $10001,700,000) (Table 1). Horses purchased at the eight yearling sales grossed $153,343,000. Exported horses A total of 356 (12.8%) horses were exported as at the 31 July 2005. Horses were exported to China (1), Hong Kong (71), Japan (7), Korea (6), Macau (37), Malaysia (69), New Zealand (74), the Philippines (10),

Australian Veterinary Journal Volume 89, No 10, October 2011

389

EQUINE

EQUINE
Table 1. Horse signalment and purchase price for each of the ve purchase price categories

Purchase price category in AUD

EQUINE

Sex at sale Colt Filly 245 (47.9%) 211 (40.3%) 347 (38.3%) 183 (36.7%) 144 (43.0%) 1130 (40.8%) Gelding 2 (0.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.1%)

Mean purchase price in AUD (median) $6621 ($7000) $15,855 ($16,000) $34,704 ($34,000) $74,372 ($72,500) $218,470 ($170,000) $55,299 ($30,000)

1. $010,000 (n = 511) 2. $10,00120,000 (n = 523) 3. $20,00150,000 (n = 906) 4. $50,001100,000 (n = 498) 5. >$100,000 (n = 335) Total (n = 2773) AUD, Australian dollars.

264 (51.7%) 312 (59.7%) 559 (61.7%) 315 (63.3%) 191 (57.0%) 1641 (59.2%)

Singapore (51), South Africa (28) and the United States of America (2). There were 18 horses that were exported from Australia and then again to another country and 31 horses that were imported back into Australia. Overall race results From the time of sale to the end of their 3-year-old race year, 2650 (95.6%) horses had been registered for racing purposes. Among the 2650 horses that were named, 2571 (97.0%) were registered with 1019 trainers during the 2-year period. Over the 2 years of racing examined, 41.7% (1071/2571) of these horses changed trainers and the average number of trainer changes per horse was 1.6 (1; 16). Of the 2773 horses that were sold, 2206 (79.6%) started in at least one race. The mean age (derived from the date of birth) at the time of their rst race was 2.8 years (2.8; 1.94.0). Overall, the mean number of race starts was 6.0 (5; 038) and for those horses that raced the mean number of race starts was 7.5 (7; 138). Overall, the mean number of places was 2.1 (1; 020) and for those horses that raced the mean number of places was 2.6 (2; 020), with 30.2% of starts per horse placed (29%; 0100%). The overall mean prize money earned was $24,420 ($2450; $03,470,325) and for those horses that raced the mean prize money earned was $30,725 ($6500; $03,470,325), with mean prize money per start of $3168 ($887; $0242,836). Over the 2-year racing period, horses earned a total of $67,594,857 in prize money. In total, 1711 (61.7%) horses earned some prize money; 312 (11.3%) earned more than $40,000, the estimated cost of training during the rst 2 years of racing; 402 (14.5%) earned prize money greater than their purchase price and 142 (5.1%) earned prize money greater than the sum of the estimated costs ($40,000) and purchase price. Association between purchase price and race performance As purchase price category increased there was an increased likelihood of horses starting in a race (P < 0.001), an increased number of race starts (P < 0.001), an increased number of race places (P < 0.001),

an increased chance of earning any prize money (P < 0.001), an increased chance of horses earning more than the $40,000 in estimated costs and a decreased chance of earning prize money greater than the purchase price in the rst 2 years of racing (Table 2). There was a signicant positive association between purchase price category and median prize money earned during the 2- and 3-year-old race seasons (P < 0.001). As purchase price increased from $0 to $100,000, the chance of a horse earning greater than the sum of the estimated costs and purchase price increased; over $100,000 the chance decreased to approximately that of horses in the $20,00150,000 category (Table 2). Multivariable analyses There was only one signicant interaction between sex (male and female) and purchase price category for any of the performance outcomes assessed and this was for the outcome of earning prize money greater than the purchase price (P = 0.04). Associations between the percentage of horses earning prize money greater than their purchase price and the purchase price category were therefore assessed separately for males and females using Fishers exact test. For both male and female horses, there was a signicant decrease in the likelihood of earning prize money greater than their purchase price with increasing purchase price category (P < 0.001), although this decrease varied with the sex of the horse (Table 3). Sex was not a potential confounder, as all four coecients of purchase price category changed by less than 15% when sex was excluded from or included in all models. After adjusting for purchase price category, males performed signicantly better than females for all race performance outcomes (P < 0.05). Discussion In this study of the Australian Thoroughbred yearling market, race performance increased with increasing yearling purchase price. With
2011 University of Melbourne Australian Veterinary Journal Australian Veterinary Association

390

Australian Veterinary Journal Volume 89, No 10, October 2011

Table 2. Number of race starts, race places, prize money earned and percentage of horses that started, earned any prize money, earned prize money greater than $40,000, their purchase price and combined estimated costs/purchase price during the rst 2 years of racing, for each purchase price category

Purchase price category in AUD >$0 >$40,000*

Starters

Mean no. of race starts (median; range)

Mean no. of race places (median; range)

Mean prize money earned (median; range)

Prize money (AUD) >Purchase price >$40,000* + purchase price 123 (24.1%) 30 (5.7%) 92 (10.2%) 347 (69.7%) 240 (71.6%) 1711 (61.7%) 92 (18.5%) 79 (23.6%) 312 (11.3%) 85 (16.3%) 117 (12.9%) 56 (11.2%) 21 (6.3%) 402 (14.5%) 14 (2.7%) 26 (5.0%) 48 (5.3%) 35 (7.0%) 19 (5.7%) 142 (5.1%)

2011 University of Melbourne Australian Veterinary Journal Australian Veterinary Association

371 (72.6%) 5.7 (5; 038) 6.3 (6; 027) 6.5 (6; 022) 6.1 (6; 020) 6.0 (5; 038) <0.001 <0.001 2.1 (1; 020) 2.5 (2; 011) 2.5 (2; 015) $45,371 (7775; 03,470,325) $56,204 (8900; 02,760,460) $24,420 (2450; 03,470,325) <0.001 2.2 (1; 013) $17,392 (3200; 0730,305) 1.8 (1; 020) $10,691 (812; 0449,914) 288 (55.1%) 595 (65.7%)

5.2 (4; 027)

1.5 (0; 013)

$9648 (0; 01,429,202)

241 (47.2%)

19 (3.7%)

396 (75.7%)

738 (81.5%)

417 (83.7%)

284 (84.8%)

1. $010,000 (n = 511) 2. $10,00120,000 (n = 523) 3. $20,00150,000 (n = 906) 4. $50,001100,000 (n = 498) 5. >$100,000 (n = 335) Overall (n = 2773) P-value <0.001

2206 (79.6%)

<0.001

<0.001

<0.001

0.030

*The estimated cost of keeping a racehorse for the 2-year period. AUD, Australian dollars

EQUINE

Australian Veterinary Journal Volume 89, No 10, October 2011

391

EQUINE

EQUINE
Table 3. Percentage of male and female horses that earned more than their purchase price during the rst 2 years of racing for each purchase price category

EQUINE

Purchase price category (AUD)

Prize money > purchase price Male Female 60/245 (24.5%) 31/211 (14.7%) 27/347 (7.8%) 14/183 (7.7%) 5/144 (3.5%) 137/1130 (12.1%) <0.001

1. $010,000 (n = 511) 2. $10,00120,000 (n = 523) 3. $20,00150,000 (n = 906) 4. $50,001100,000 (n = 498) 5. >$100,000 (n = 335) Overall (n = 2773) P-value AUD, Australian dollars.

63/266 (23.7%) 54/312 (17.3%) 90/559 (16.1%) 42/315 (13.3%) 16/191 (8.4%) 267/1643 (16.3%) <0.001

The results of this study of Australian yearling Thoroughbreds are consistent with those from studies of other Thoroughbred yearling markets, which show that owning Thoroughbred racehorses tends to be a losing proposition. Almost 40% of horses followed in this study earned no prize money as 2- and 3-year-olds. In a similar study of a regional Thoroughbred yearling market in Canada, Robbins and Kennedy11 found that 30.6% of horses earned no money; 64% of 537 2-year-old horses in training in the UK17 and 39% of 1567 2-year-old horses followed in Australia16 earned no money. Overall, in this study approximately 11% of horses earned more than the estimated cost of training during the 2 years following purchase, which is consistent with the ndings of others. In a study aimed at proling the racing career of Thoroughbred horses in Australia, More reported that 13% of 2-year-old horses recouped their training costs.16 Only 5% of 2-year-olds and 17% of 3-year-olds were able to earn back the estimated costs of training in a study investigating wastage in Thoroughbred horses in the UK.17 We used a at cost of $20,000 per year for training fees, based on surveying the monthly advertised costs for a mean-priced horse being syndicated in 2005. It is probable that higher priced yearlings have higher training fees than lower priced yearlings; however, these data were not available. The mean training fee for horses trained overseas may also dier from the mean training fee for horses trained in Australia. Only 5.1% of horses in the present study earned more than the combined purchase price and estimated training costs within the 2-year racing period, which is consistent with the ndings of Robbins and Kennedy,11 who found that only 4.6% of 1316 Canadian yearlings sold were likely to be protable in the rst 2 years of racing. As others have found,1618 there was considerable wastage (failure to race) in the early stage of a Thoroughbreds racing career, with approximately one in ve horses failing to start in a race. The wastage was greatest for horses purchased for $10,000 or less, and least among those horses purchased for more than $100,000. These results may indicate that more expensive horses have superior athletic ability and are therefore more likely to race or could simply imply that purchasers paying higher prices are more willing to persist with getting their purchase to the racetrack. Kane et al. in a study of 1162 Thoroughbred yearlings sold in the USA reported a similar proportion of horses failing to race at 2 or 3 years of age.18 In a UK study assessing causes of wastage in Thoroughbred training establishments, 39% of 537 2-yearolds and 24% of 456 3-year-olds in training failed to race.17 The overall observed poor return from purchasing Australian Thoroughbred yearlings, the observed positive association between purchase price and subsequent race performance and the decrease in returns relative to investment with increased race performance is very similar to results reported for the United States market. Studies that have estimated the rate of return from ownership of Thoroughbreds in the United States conclude that purchasers of Thoroughbred yearlings pay two types of premiums to own a racehorse.13,14 First, all owners pay a premium to enter the sport of racing. That premium corresponds to the dierence between the poor rate of return from investment in yearlings and the rate of return available from investment in other nancial assets of comparable risk. Presumably, owners
2011 University of Melbourne Australian Veterinary Journal Australian Veterinary Association

increasing purchase price there was an increase in the proportion of horses racing and a signicant dierence in the distribution of the number of race starts and places, and the prize money earned. This result is consistent with the ndings of Heckerman,15 who collected data from yearling sales in the United States from 1980 to 1993 and showed that with increasing purchase price there was an increase in the proportion of horses winning stakes races and the mean prize money earned. It can therefore be considered benecial to include purchase price in any modelling of race performance.18 The most obvious explanation for the observed positive association between the purchase price of yearlings and subsequent racing performance is that the market has an ability to identify superior horses, but other reasons for the observed relationship are also possible. For example, a greater initial investment in a horse is likely to be associated with a greater ongoing investment, such as training fees, which may also result in superior racing performance, as well as greater motivation to pursue a return. The purpose of this study was to describe the relationship between purchase price and a number of race performance outcomes. It is possible that there are a number of factors, such as training, diet, genetics and conformation, that lead to the observed relationship. Quantifying the relative contribution of these factors to the observed relationship is a matter for subsequent studies. Although paying more money for a yearling was associated with an increase in mean earnings, in the highest quintile the increase was insucient to cover the cost of the increased investment in these horses. For example, the total cost of horses in category 1 ( $10,000) was $3,383,750, and these horses earned $1,507,392 more than their total cost. In contrast, horses purchased for more than $100,000 (category 5) earned $54,359,121 less than their total cost of $73,187,500.

392

Australian Veterinary Journal Volume 89, No 10, October 2011

EQUINE
obtain non-nancial benets from participating in the sport. Second, owners who wish to invest in high-value yearlings pay an additional premium to own a prospective champion. This premium corresponds to the dierence in the poorer rate of return for the highest priced yearlings and the rate of return available from lower priced yearlings. Price is no object when you are going for the best becomes a signicant driver of such markets.14 Although this study did not estimate rate of return, our data strongly indicate that both premiums also exist in the Australian Thoroughbred yearling market. This study only analysed performance in the rst 2 years of racing and did not consider potential future earnings from racing, breeding or on-sale. The expected residual value of the horse once its racing career is completed is probably an important factor in determining the purchase price. In this study, we observed that the proportion of males gelded decreased with increasing purchase price category, which may indicate that owners purchasing colts of high cost were hesitant to geld a potential future stallion prospect. However, DeGennaro included residual values in his modelling of the United States Thoroughbred market and still found negative cumulative rates of return and that the rates of return, even with residual value of horses included, were least attractive for the most expensive horses.14 Conclusions The purchase price of Australian Thoroughbred yearlings was found to be positively associated with the number of race starts, number of race places and prize money earned. Overall, only 5.1% of horses earned more than the combined purchase price and estimated training costs within the 2-year racing period. This study provides strong evidence that the Thoroughbred yearling market in Australia behaves in a similar manner to the United States market in which owners pay a premium to enter the sport of racing and an additional premium in the quest to own a champion horse. Researchers examining the race performance of yearlings purchased at sales should consider including purchase price when modelling. Acknowledgments This study was funded by the Rural Industries Research and Development Corporation and Racing Victoria Ltd. The cooperation of Magic Millions Pty Ltd and William Inglis and Son Ltd is gratefully acknowledged. The assistance of Mr Ito, Dr Cust, Mr Webbey, Mr Magboo, Mr Wanklin and Ms Neal in obtaining overseas racing information is very much appreciated. References
1. Australian Racing Board Ltd. Size and scope of the Australian Thoroughbred racing industry. Australian Racing Board Ltd, Mascot, 2001. 2. Gordon J. The horse industry: contributing to the Australian economy. Rural Industries Research and Development Corporation, Canberra, 2001. 3. Ross J. Australian racing fact book 20022003: a guide to the Australian racing industry in Australia. Australian Racing Board Ltd, Mascot, 2003. 4. Preston SA, Zimmel DN, Chmielewski TL et al. Prevalence of various presale radiographic ndings and association of ndings with sales price in Thoroughbred yearlings sold in Kentucky. J Am Vet Med Assoc 2010;236:440445. 5. Buzby JC, Jessup EL. The relative impact of macroeconomic and yearlingspecic variables in determining Thoroughbred yearling price. Appl Econ 1994;26:18. 6. Chezum B, Wimmer B. Roses or lemons: Adverse selection in the market for thoroughbred yearlings. Review Econ Stat 1997;79:521526. 7. Hastings T. Price variations of yearling thoroughbreds at Australian auctions. Irish J Agric Econ Rural Sociol 1987;12:1928. 8. Karungu P, Reed M, Tvedt D. Macroeconomic factors and the thoroughbred industry. J Agric Appl Econ 1993;25:165173. 9. Neibergs JS, Thalheimer R. Price expectations and supply response in the Thoroughbred yearling market. J Agric Appl Econ 1997;29:419435. 10. Neibergs JS, Thalheimer R. An economic analysis of the eectiveness of Thoroughbred breeder-owner incentive policies. J Agric Appl Econ 1999;31:581592. 11. Robbins M, Kennedy PE. Buyer behaviour in a regional thoroughbred yearling market. Appl Econ 2001;33:969977. 12. Parsons C, Smith I. The price of thoroughbred yearlings in Britain. J Sports Econ 2008;9:4366. 13. Gamrat FA, Sauer RD. The utility of sport and returns to ownership: evidence from the Thoroughbred market. J Sports Econ 2000;1:219235. 14. DeGennaro RP. The utility of sport and returns to ownership. J Sports Econ 2003;4:145153. 15. Heckerman DL. Stakes winners in abundance. Blood-Horse 1996;32163220. 16. More SJ. A longitudinal study of racing thoroughbreds: performance during the rst years of racing. Aust Vet J 1999;77:105112. 17. Wilsher S, Allen WR, Wood JLN. Factors associated with failure of Thoroughbred horses to train and race. Equine Vet J 2006;38:113118. 18. Kane AJ, McIlwraith CW, Park RD et al. Radiographic changes in Thoroughbred yearlings. Part 2: Associations with racing performance. Equine Vet J 2003;35:366 374. 19. Spike-Pierce DL, Bramlage LR. Correlation of racing performance with radiographic changes in the proximal sesamoid bones of 487 Thoroughbred yearlings. Equine Vet J 2003;35:350353. 20. Cohen ND, Carter GK, Watkins JP, OConor MS. Association of racing performance with specic abnormal radiographic ndings in Thoroughbred yearlings sold in Texas. J Equine Vet Sci 2006;26:462474. 21. Australian Jockey Club Ltd, Victoria Racing Club Ltd. The Australian stud book. 2005. http://www.studbook.org.au. Accessed February 2004 to August 2005. 22. New Zealand Thoroughbred Racing Inc. The New Zealand stud book. 2005. http://www.nzracing.co.nz/Breeding/Studbook.aspx. Accessed May-August 2005. 23. William Inglis and Son Ltd. [Sales]. http://www.inglis.com.au/sales/results/. 2004. Accessed August 2003 to January 2004. 24. Magic Millions Pty Ltd. [Online catalogues, sales archives]. 2004. http:// www.magicmillions.com.au/sales_info_archives.html. Accessed August 2003 to January 2004. 25. Racing Information Services Australia Pty Ltd. i-ris database. 2005. http:// www.racingservicesbureau.net.au/roarrs. Accessed August 2004 to December 2005. 26. The Hong Kong Jockey Club. Racing results for Hong Kong. 2005. http:// www.hkjc.com/english/hrc/hindex.asp. Accessed NovemberDecember 2005. 27. New Zealand Thoroughbred Racing Inc. New Zealand racing results. 2005.http://www.nzracing.co.nz/Meetings/HorseSearch.aspx. Accessed AugustDecember 2005. 28. Korean Racing Authority. Korean stud book and racing results. 2006. http:// studbook.kra.co.kr/eng/main.jsp. Accessed January 2006. 29. Beijing Jockey Club. Beijing horse racing results and stud book. 2005. http:// www.mawww.com/defaulten.asp. Accessed November 2005. 30. Tiago Stock Consulting. Currency converter. 2006. www.x-rates.com/. Accessed May 2006. 31. Ken King Thoroughbreds. Horses for syndication. 2005. http://www1. kenkingthoroughbreds.com.au/. Accessed October 2005. 32. Hancox Bloodstock Pty Ltd. Hancox bloodstock: new syndicates. 2005. http:// www.hancox.com.au/index.html. Accessed October 2005. 33. Delbridge Racing. Horses available. 2005. http://www.delbridgeracing. com.au. Accessed October 2005. 34. Hilbe J. Negative binomial regression. Cambridge University Press, Cambridge, 2007. 35. Dohoo I, Martin W, Stryhn H. Veterinary epidemiologic research. VER Inc., Charlottetown, Prince Edward Island, 2010. (Accepted for publication 20 March 2011)

2011 University of Melbourne Australian Veterinary Journal Australian Veterinary Association

Australian Veterinary Journal Volume 89, No 10, October 2011

393

EQUINE