Is the level of concentration in Irish beef processing lowering cattle prices? : an empirical approach

(1)

1

Is the level of concentration in Irish beef processing lowering cattle prices? :

An empirical approach

Donnacha Hennessy

10824332

Abstract

In light of growing concerns regarding the nature of competition in the market for cattle in Ireland, this thesis examines the relationship between concentration in beef processing and cattle prices. By applying Price Concentration Analysis and Entry and Exit analysis to a unique dataset, compiled by the author, this research tests this relationship. I find that with greater concentration of beef factory ownership in the hands of the “Big Three” beef processors, the price per kilo paid for cattle at Irish beef factories falls. The recently cleared merger Joint Venture of ABP/Fayne Valley Group is discussed in light of these findings.

University of Amsterdam

Faculty of Economics and Business

MSc. Economics – Specialisation in Markets and Regulation Masters Thesis

Supervisor: Dr. Jo Seldeslachts Date: 24th_{of May 2017}

(2)

2

1. Introduction

Irish beef farmers have long complained of the anti-competitive effect of the level of concentration in the beef processing sector, which is dominated by the “Big Three” food groups, ABP, Keepak and Dawn. Together the Big Three’s share of the cattle slaughtered in Ireland is reported as being higher than 70% (McCloughan, 2016). This has become a politically divisive issue, with farmers regularly protesting and even blockading factories (Irish examiner, 2014).

This issue has been topical in recent months. ABP, the largest processor in Ireland recently entered into a joint venture with Fayne Valley Group to jointly operate Irish Country Meats (“ICM”) a slaughterhouse with a national share of 8% of live cattle purchased for slaughter. On the 7th_{of October the EC cleared} unconditionally the acquisition of 50% of Slaney Foods (owner of ICM) by ABP Group. (EC Press release, 2016) This merger was heavily protested by Irish farmers, who believed that this would harm

competition in this market.

The theories of harm correspond to the economic theories of unilateral and coordinated horizontal effects. To date most of the research on the competitiveness of the sector has focused on the claims that beef processors have colluded, lowering prices for live cattle artificially (Rogers, 2004). Economic analysis and legal investigations have so far uncovered no evidence of a cartel operating (Macconnell, 2000). However whether or not the level of concentration has affected the level of competition in this market is unexamined.

Ireland is the largest exporter of beef in Europe, with exports approximately €2.41 billion, a large sum to a small economy (Bordbia, 2016). Over 100,000 farms have a beef enterprise, and Department of Agriculture, Food and Marine (“DAFM”) approved export plants provide employment of between 12,000 to 15,000 workers, making this sector a major source of rural employment1_{. Ensuring that this sector is} functioning well is therefore a matter of great importance.

The importance of this sector, combined with the gap in the literature surrounding the state of competition in the sector motivate the following research questions.

i) Is the level of concentration in the beef processing sector leading to lower prices in the market for live cattle?

ii) Should the Linden/Fayne Valley Joint Venture be expected to lead to lower prices in the market for live animals?

A unique dataset compiled by the author is used to answer these questions. Applying econometric methods, I find that when the concentration in ownership of plant factories increases the price paid to a farmer for cattle falls. The results of this study have implications for Irish industrial policy.

The remainder of the paper is structured as follows. Section 2 provides background information and a review of the relevant literature on oligopsonistic competition and the Irish market in order to frame my

1_{While it is difficult to put an exact figure on the number of people employed in the Irish beef sector, DAFM puts the figure of Irish farms with a} beef enterprise at 100,000, with 68,300 being Beef specialists. Adding to this are the thirty large processing plants which operate in Ireland. These plants typically employing 200-250 people directly, and a further 200-250 indirectly (AgriAware, 2013).

(4)

4 research. Section 3 introduces my dataset and methodology. Section 4 presents empirical results. Section 5 gives my findings and conclusions.

(5)

5

2. Literature Review

In this section, I review the literature to relating to a) oligopsony power, and b) whether there is oligopsony power in the Irish market for cattle for live slaughtering

2.1 Oligopsony power

Here I review the literature regarding the existence of oligopsony power in agricultural markets Sexton (2012) points out that few agricultural markets possess the main characteristics of perfect competition, many small producers and buyers, perfect information, and homogenous goods. Many agricultural markets are instead characterized by oligopsony (Chen & Lent, 1992). This model matches closely with primary agricultural markets in particular, where typically a few large processors are supplied by a large number of small farmers.

2.1.1 Buyer Concentration

These markets typically exhibit high levels of buyer concentration. Crespi , Richard and Sexton (2012) found that for selected meatpacking industries, the combined market share of the largest four

purchasers (CR4) in 2007 was 80%, 70%, 65%, 57%, and 51%, respectively, for steers and heifers, sheep and lambs, hogs, broilers, and turkeys. Unfortunately, the available information on European markets is less recent. It is clear that even as early as the 1990’s such markets demonstrated high buyer

concentration, higher even than that in US (Sheldon, 2017).

Concentration appears to be increasing in the industry. Crespi , Richard and Sexton (2012) found evidence that concentration increased in these markets over the period 1980 to 2010 the CR4 in these meatpacking industries increased on average by 69%. Unlike the US in Europe most research has focused on the effect of concentration in the downstream market amongst retailers, with little focus on concentration in upstream markets (Sheldon, 2017).

2.1.2 Buyer Power

This high level of concentration in buyers can result in buyer power. Chen (2007) describes buyer power as;

“the ability of a buyer to reduce the price profitably below a supplier's normal selling price, or more generally the ability to obtain trade terms more favourable than a

supplier's normal trade terms” (p.19).

Dobson, Clarke and Davies (2001) find that for buyer power to be present the buyers purchases constitute a substantial portion of purchases in the market; supply curves are upward sloping, and barriers to entry into the buyer's market exist. All of these criteria are present in primary agricultural markets, given that a few large buyers dominate purchasing, the quantity supply increases with price and entry involves high sunk costs.

Mergers by their very nature increase concentration, and can therefore increase buyer power (Thomas & Fee, 2004). Bhattacharyyaa and Nain (2011) have gone further, demonstrating that mergers do increase buyer power by showing that input suppliers’ prices and operating profits typically fall post-merger.

(6)

6 Competition analysis typically views buyer power as beneficial, but it can also result in inefficiencies. Chen (2007) draws the distinction between counterveiling buyer power and oligopistic power, two types of buyer power which have contrasting effects on welfare. It is the level of concentration in purchasing that differentiates between these two concepts. Counterveiling buyer power occurs when suppliers are also few and large, and hold market power. In this case buyer power can have beneficial effects as it allows buyers to counter a market power. For this reason it is generally thought to be efficiency enhancing.

Chen (2007) states that oligopsonistic power occurs when suppliers are small and many, and supply is therefore competitive. In this case there is no producer market power to mitigate, and a reduction of the selling price is a move away from the “normal selling price”. As the oligopsonist purchases a high share of total purchases, the oligopsonists’ consumption can therefore influence the price. The marginal expenditure for the oligopsonist is therefore comprised of both the marginal cost of the good, and the change in prices due to a moving up or down the supply curve. As the relationship between price and quantity supplied is positive, the marginal expenditure curve is always higher for a given quantity than the supply curve. This results in an oligopsonist consuming the quantity 𝑍𝑚, which equates the marginal expenditure to its willingness to pay. Beyond this point the cost of an additional unit exceeds the value the oligopsonist places on it.

Therefore both the quantity and price are now lower than at the socially optimal equilibrium, which would be achieved by perfect competition. This results in an allocative inefficiency as producers that would be willing to produce for prices between 𝑊𝑀 and 𝑊∗do not, despite the fact that the

oligopsonists willingness to buy exceed the cost of these units. This deadweight loss is represented in the triangle between 𝐴, 𝐵 and 𝐶 in Figure 1.

Figure 1: Oligopsony model

For a small departure from competition, the deadweight loss triangle is small. But if market power is severe, or is exercised at multiple stages along the market chain, deadweight losses become large. Sexton and Zhang (2001), found that even small deviations from perfect competition in the vertical

(7)

7 market this can lead to a loss of 25% of the total market welfare. Therefore even a small amount of oligopsony power being exercised can cause large inefficiencies across the supply chain if downstream markets are also characterized by imperfect competition.

2.1.3 Does oligopsony power necessarily lead to an efficiency loss?

Even if oligopsony power exists, a processor may not have an incentive to reduce their purchase of cattle. Processors may want to run their plants at a high capacity, in order to reduce average total costs. This would prevent a processor from reducing the quantity it demands (Crespi, Sexton & Saitone, 2012). Processors have contractual obligations to meet demand in the downstream market. Thus in effect their demand may be set by retailers, as not meeting these obligations entails significant reputational and financial costs (Sexton, 2012; Crespi, Sexton & Saitone, 2012). Furthermore with contracts and non-linear pricing this efficiency can be eliminated (Chen, 2007). This is now seen in the US, where markets are becoming more vertically integrated, and on the spot markets less common.

2.1.4 Evidence of oligopsony power

Given the high level of concentration in primary agricultural markets it is no surprise that buyer power has been identified empirically in many such markets. Xia and Sancewich (2012) listed the following agricultural markets in which buyer power has been identified includes; Wheat (Stiegert & Hamilton, 1998), Cocoa (Wilcox & Abbott, 2006), Tobacco (Raper, Love, & Shumway, 2006), Tomatoes (Huang & Sexton, 1996), Hogs (Zheng & Vukina, 2009) and Milk (Alvarez, Fidalgo, Sexton & Zhang, 2000). Several studies that have examined oligopsony power in markets for live cattles. Both Azzam and Anderson (1996), and more recently Ward (2002) conducted a review of studies that examined the relationship between processor concentration and live cattle prices. Azzam and Anderson (1996) found that while many studies did find that increase processor concentration2_{resulted in lower prices at an} aggregate or feedlot level, and the size of the effect was typically quite small (<5%). Ward (2002) also found that higher concentration3_{resulted in lower prices for live cattle in the majority of papers that} investigated this link. More recent research, by Love, Capps and Williams (2009) found that beef processors continue to pay less for fed cattle in more concentrated regions.

2.2 The Irish market for live cattle

In this section I describe the features of the Irish market which would contribute to such oligopsony power

2.2.1 Overview

The market for live cattle is supplied by animals which originate on “suckler” farms. These farms consist of a herd of cows which birth a calf a year. These animals are then fattened either on this farm or “finished”, or fattened at another farm, for various lengths of time. Finally the animal is sold to beef factory, before being transported to the slaughterhouse for slaughter. This stage marks the end of the market for live cattle for slaughter.

2_{In these papers the measures of competition used were market share based CR3, CR4 or Herfindhal-Hirschman Index (“HHI”)} 3_{Market share based CR3, CR4 or HHI, number of bidders or entry or exit of a factory}

(8)

8

2.2.1 Demand

Beef factories are typically owned by large food groups which own multiple factories, with the remainder run independently as a competitive fringe. Table 1 shows the level of concentration in the market for live cattle for slaughter in Ireland. As can be seen below, the three largest processors, ABP Foods, Dawn and Keepak, slaughter the majority of live cattle in Ireland. The CR3 in this market at roughly 70%, is high (McCloughan, 2016). Concentration in processing capacity mirrors this, with the Big Three owning over half of all factories.

Table 1: DAFM approved export plants

Group ROI Market Shares by volume in 2015 Number of Plants ABP 30-40% 7 Dawn 10-20% 3 Kepak 10-20% 4 CR3 c. 60-70% 14 Other processors c. 30% 13 Total 100% 25

Source: Market shares from ABP/Foyne Valley merger decision

If the geographic scope of the market is sub-national, we can see from Figure 1 that the CR3 would be even higher in the south or south western markets.

(9)

9

Figure 2. Map of beef factory ownership

Source: locations found online by author

The Big Three faces little competition from competitors other than DAFM approved export plants. Exports are small with the vast majority of cattle sold for live slaughter in Ireland being purchased and slaughtered by DAFM approved beef factories, representing over 1,600,000 kills (DAFM, 2016). Smaller local authority plants in contrast slaughter less than 100,000 cattle a year combined, less than 6% of the domestic consumption, and therefore impose a small competitive constraint (DAFM, 2016).

The Big Three therefore purchase a large share of the total purchase of cattle, the first prerequisite of buyer power (Chen, 2007). Entry in this market requires large investment in facilities that are largely sunk costs. Hence high barriers to entry, the second requirement, is also present.

2.2.2 Supply

The third and final requirement for buyer power is an upward sloping supply curve. This is what allows a factory to push down prices by decreasing the quantity it consumes (Chen, 2007). The supply curve for cattle is upward sloping in the short run as it can rise or fall quickly in response to price changes. Farmers can withhold cattle in the face of a low price or send cattle they would otherwise continue to fatten in response to a higher price.

In order for such buyer power to constitute oligopsony power, it must be held in the face of a

competitive supply (Chen, 2007). There are over 60,000 farms which specialize in beef production, with many more which produce cattle alongside other animals or crops (Department of Agriculture, 2009). Although some farmers collectively sell to factories through selling groups, the majority of animals are purchased directly from independent farmers. Irish beef producers then clearly lack any significant

(10)

10 market power, being both numerous and small. Buyer power held by beef processors is therefore not counterveiling buyer power, but oligopsony power (Chen, 2007).

Irish beef producers are dispersed throughout the island, being present in every county. As Figure 3 shows, there are higher concentrations in the Midlands, North Leinster, and Munster, and far fewer in Mid-Leinster and on the west coast.

Figure 3: Map of beef production

Source: DAFM 2016

2.2.3 Do Irish processors have an incentive to exercise oligopsony power?

As discussed earlier, when assessing the likelihood of oligopsony power resulting in an efficiency loss, context is of great importance. The mere existence of oligopsony power does not lead to an efficiency loss, as processors may not have sufficient incentive to lower the market price. I will here review whether or not the factors the literature review indicated could dampen this incentive are likely to influence a beef processor in Ireland.

In Ireland most factories already operate far below capacity, with many operating only three days a week (McCloughan, 2016). This may be evidence of oligopsony power, as it is through decreasing

(11)

11 quantity that the price decrease is achieved. If not, it is still a sign that average costs are far from

minimized, and a likely unprofitable pursuit. Therefore the Big Three are unlikely to forego excercising oligopsony power in an effort to improve cost efficiency.

Unlike the US, in which pricing is more sophisticated (Crespi, Sexton & Saitone, 2012) in Ireland cattle are not purchased through contracts. In Ireland a farmer selling cattle contacts factories directly for on-the-spot quotes for the weekly price per-kg, or can check DAFM’s PriceWatch App. Therefore non-linear pricing has not resolved the efficiency loss.

(12)

12

3. Empirical Approach

In this section I describe the dataset that was compiled, and the econometric methods which is used to test the relationship between concentration and prices

3.1 Data

This analysis relies on a unique dataset which was compiled from a variety of sources. Weekly data on beef prices for all large slaughterhouses is published by the DAFM, and is freely available online. The weekly average of the price paid per Kilogram at each slaughterhouse for different cattle types is available in these reports. From this a monthly average was calculated by taking an average of all weeks beginning in that month. Quarterly data on the percentage breakdown of each classification

combination is also available at a slaughterhouse level. The percentage of cattle graded lower in fat, and graded better in conformation, is taken from these reports.

Addresses and ownership details of all listed factories was taken from their websites. The geographic coordinates of factories and ports were established from searching Google Maps. Distances between plants and ports were calculated using the Geographic Distance Matrix Calculator. This is a free, Java application that can compute all pair wise distances from a simple list of coordinates (Centre for Biodiversity, 2016). Currency information was taken from XE.com.

This data captures one slaughterhouse entering and one slaughterhouse leaving the market. It is this entry and exit which gives the measure of competition within plant variation. Ballon Meats, an

independent slaughterhouse located in the South East, began operations in February 2013. Liffey Meats Ballinasloe, owned by Liffey Meats and located in the Mid-West left the market in January 2015. Neither were operated by the Big Three. At the radius of 200km the entry of Ballon affected the concentration ratio of all but four factories, and the exit of Liffey Meats effected all but two. At the 300km radius Ballon Meats entry and Liffey Meats exit effected effected the CR3 of all factories. At these distances the CR3 of almost all factories is affected. Each incident changed the CR3 in affected plants ratio by 4-5% in these geographic markets.

3.2 Methodology

As previously noted, the effect of oligopsony power on prices has been investigated in many other live cattle markets (Azzam & Anderson, 1996; Ward, 2002; Love, Capps & Williams, 2009). In many of these studies data on quantities and costs were required. This information is not publicly available for the Irish market, and I therefore utilize two econometric techniques which are feasible given the data available to me. These are Price Concentration Analysis (PCA) and Entry and Entry and Exit Analysis (EEA).

These techniques are frequently used by competition authorities to investigate the link between market structure and measures of performance. In the following passages I outline both techniques.

2.3.1 Price Concentration Analysis

Price Concentration Analysis (PCA) involves the use of econometric techniques to estimate the relationship that exists between changes in market structure and a performance measure, usually

(13)

13 prices. PCA can even quantify this increase. This has made PCA a popular means for competition

authorities the effect of a change in market structure on prices in mergers and market investigation4_. In order to examine the relationship between concentration and prices, the product and geographic market must first be defined. This allows for the accurate measurement of the measure of the three ingredients necessary to conduct PCA (Wirth & Akinrinade, 2014). These are:

 The measure of competition  The measure of performance

 Methods for controlling for endogeneity

Market definition

Irish beef factories, like those in continental Europe produce for supermarkets for whom different animal meats are not substitutable. There is no therefore no reason to stray from the European

Commission’s definition of the relevant product market in previous, that of the “purchasing of live cattle for slaughtering” (European Commission, 2016,).

Geographic market definition is a more complicated issue. The Competition Authority (TCA) found that over 90% of cattle slaughtered at a beef factory were purchased within a 60km catchment area (The Competition Authority, 2003). In a more recent determination the catchment area appears to have increased with the EC has found that cattle are rarely transported over 100km(European Commission, 2016). This would support 100km as the narrowest possible market definition.

As the TCA noted in decision at the time, prices typically vary little between areas for any market definition suggesting that overlapping catchment areas have resulted in a “chain of substitution effect”. This means that a factories can exercise a competitive constraint even on factories with which it does not compete with for the same cattle. Therefore factory A can constrain factory C, despite not

competing for the same farmers cattle, as it competes with factory B, which does compete with factory C. A, by imposing a competitive constraint on B’s ability to increase prices in response to a SSNIP by factory C, is indirectly imposing a constraint on C.

This makes defining the upper limit of the Irish market(s) for the laughter of live cattle challenging. As no clear geographic market definition exists, the market definition was initially open in this analysis, though not all results are presented here. The map of slaughterhouse level prices over time shows little

variation across factories, with prices of most factories in most periods falling in a 3c per kg band. This implies that the geographic extent of the market for cattle for live slaughter is national. Were the country divided in more regional markets it should be expected groups of nearby factories to be banded together, possibly creating some sort of geographic incline (east-west or north-south). As each factories 300km radius captures most other slaughterhouse, this can be treated as an almost national market.

4_{The CC made frequent use of it in cases such as Anglo American/Lafarge (2012), Cineworld/City Screen (2013). In TNT/UPS (2012)}4_{it was its} PCA that convinced the European Commission that the anti-competitive effects of the merger would outweigh any potential efficiencies.

(14)

14

Figure 4: Beef factory prices over time

Measure of Competition

A wide variety of measures of competition have been used in PCA to date. These are typically based on market shares by volume, value or capacity, the number of competitors, or the presence of specific competitors. The market share which best captures competition varies between markets. In markets with differentiated goods, market shares by value more accurately capture market power than those based on volume. In markets where it is a firms’ capacity that determines its market power, a firms’ share of capacity is more informative (The Office of Fair Trade, 2010).

Using firms purchasing shares it is possible to measure the buyer power held by firms within a market. Where a small number of firms wield most of the market or buyer power, concentration ratios can be used to measure the market power that is possessed by the large firms in this market. A concentration ratio is the sum of the market shares of the largest firms in a market, typically the largest three or four. This measure better captures market power in markets where the strength of a few large operators matters most (The Office of Fair Trade, 2010).

Information on the quantity of cattle purchased at individual factories is collected by the Department of Agriculture. Due to commercial sensitivity this data is not publicly available. Therefore it is not possible to calculate measures of competition based on volumes or revenues.

The measure of competition I use is therefore capacity based. In the market for live slaughtering it is possible to capture the market power held by groups with a measure of capacity for two primary reasons. These are that all slaughterhouse possess some spare capacity, and that factories compete for cattle along two parameters, price and proximity. A concentration in slaughterhouse ownership

therefore will reduce the incentive for factories that could impose a competitive constraint, due to the 300 350 400 450 500 Price per Kg

(15)

15 higher level of overlapping catchment areas. This is then amplified by the chain of substitution effect. This means that the market power seen in a market is proportionate to the overlapping and interlinking of catchment areas.

Of the more well-known capacity based measures that are calculable (testing the effect on price of the presence of a specific competitor, or counts of plants or fascia), most are ill suited to capturing market power in this market. I rate any measure of competition by the following three criteria;

1. The measure must be viable given the limitations of the data. 2. Theoretically sound.

3. It must be easily interpretable.

It is possible to calculate all the commonly known capacity-based measures. Furthermore as evidenced by their use in competition authorities analysis these have a strong theoretical underpinning. As

explained earlier ceretibus paribus more competitors means a more competitive outcome, in this case a higher price for farmers.

In the market for live cattle for slaughter however, interpreting these factors is impossible. Testing the presence is meaningless as in almost all cases every competitor is present in every market, due to the likely wide breadth of the geographic market. Due to the large number of independent factories, the plant count for most geographic definitions of each market is above ten. How informative is it to know the impact of the eleventh processor? One solution is to drop those smaller independent operators, as was done inthe Competition Commissions’ Aggregates Market Investigation (Competition Commission, 2011). Unlike Aggregates however capacity does not constrain these plants ability to complete a transaction, and nearby independents could impose more of a competitive constraint than a group slaughterhouse at a greater distance.

The measure of competition calculated here is the concentration of slaughterhouse ownership within each factories geographic market. This measure, the percentage of factories, is to the best of the authors’ knowledge, novel. This measure is calculable given the available data, as factories group ownership is available online. Furthermore this measure has the same theoretical underpinning as market share based measures. That is as firms are differentiated spatially, the more differentiated a slaughterhouse is, the more able it is to increase prices. Intuitively farmers have more options when selling their cattle in a market with a lower concentration of ownership, due to the higher number of groups. Conversely in a more concentrated market, where most slaughterhouse are owned by ABP, Dawn, and Keepak, farmers have fewer alternative buyers, and are therefore less likely to switch in response to lower prices.

This approach treats all individual factories as having the same competitive effect. All factories can be assumed to impose a competitive constraint through their spare capacity. This measure however does not differentiate between those factories nearest and furthest from the slaughterhouse. As nearest factories should impose a stronger constraint, this is a simplification.

Concentration is measured by the concentration ratio of the three largest food groups, ABP, Dawn and Keepak. This is the sum of the share of factories owned by these groups within a given market. This measure gives easily interpretable results as its coefficient gives an estimate of the increase in price expected for an increase in slaughterhouse ownership.

(16)

16

Measure of Performance

Despite the name, other measures of performance such as margins can be used in PCA to assess the performance of a firm in a market (Hackl et. al, 2014). However prices are preferred in practice due to the ease of their interpretation, calculation and availability.

From DAFMs weekly price reports, the average monthly price of an R3 Steer, the most typical cow sold at factories, was calculated at a slaughterhouse level. This was done by averaging the weekly price quoted by all reports beginning in that month. Steers share of the national kill was above 35% in every year, and has remained relatively stable (DAFM, 2016). While not all factories slaughter every sex category, all purchase steers.

R3 is the most typical classification of steer purchased. Factories use an R3 as the base rate from which they estimate a Steers price per kilo, adding and subtracting as steers vary from this grade. Therefore the price of an R3 Steer is likely the best measure of price, and therefore performance, in this market.

Controlling for endogeneity

The presence of unobservable omitted variables which can influence both market structure and prices can result in biased estimators (Stock & Watson, 2007). Even in distinct markets with shared or correlated demand or cost factors, it is likely wrong to assume that no differences exist. Using Panel Data, the fixed effects model can be employed to control for differences between groups (Stock & Watson, 2007). The coefficient for market structure is then estimated using only within group variation in prices and concentration, and time invariant differences between groups cannot bias the coefficient of interest (Stock & Watson, 2007). Fixed effects is particularly likely to remove all omitted variable bias when the measure of competition is structural, as it is unlikely to correlate with time variant factors (Stock & Watson, 2007).

This approach makes no use of all the variation in prices and structure between markets, and therefore this technique should not be used blindly. Fixed Effects approach is warranted for solving endogeneity if the three following criteria are met (Garces and Davis, 2009).

i) There is only limited or no data available on the drivers of unobserved differences ii) It can be assumed that the true causal effects are common across groups.

iii) There is reason to suspect that unobserved factors are likely to determine y.

Should these hold, it is then likely that the results will be biased, without the use of fixed effects. It is likely that differences exist between markets, which are hard to capture, and the causal effects can be expected to be the same across groups. More weight should be given to the fixed effects than the OLS (Garces and Davis, 2009).

There are two omitted variables in particular that could lead to endogeneity. By choosing to locate near to both ports and areas of high bovine density, a factory operator can lower his transport costs for both the purchasing of live animals, and the exporting of finished cuts. This explains why such a high

proportion of the factories are located in areas with very high bovine density, as seen in Figure 2, or on the southern or eastern coast. Many independent factories operate only in areas of high bovine density near the coast. Factories in these areas could potentially offer farmers a higher price, having lower (transport) costs. However as the independent factories bunch together in these areas, the CR3 of these

(17)

17 factories in these areas is lower than elsewhere. Therefore supply and proximity to ports can influence both prices and the measure of concentration, CR3. As these factors would simultaneously decrease the CR3 and increase prices, it would dampen the true effect of concentration on prices. A fixed effects regression, by only using variation within factories could account for this potential source of endogeneity.

Entry is not exogenous, but influenced by the profitability of entering a market. Therefore there are possibly unobserved factors that could influence both a firms entry decision, and prices (Mazzeo, 2002). It has been demonstrated that this bias can occur when the potential correlation between the error term and the competition measure is not corrected (Manzouk & Moul, 2008). Where possible an

Instrumental Variable approach is therefore preferable to a fixed effects, as this isolates the effect of the exogenous variation in the measure of competition on prices.

Finding a good instrument is always a challenging task and unfortunately several measures that were identified as candidates instrumentation were unsuitable. In this dataset entry and exit are limited however and therefore bias is unlikely to be large. Therefore I used a fixed effects regression in an attempt to control for some endogeneity. There is only limited or no data available on the drivers of unobserved cost differences between plants. Prices are clearly set to some degree by the costs a factory has in processing the cattle it buys. Theory suggests that the causal relationship between concentration and prices, and this would be true across groups. Therefore the three criteria set by Davis and Garces (2009) are met, and therefore the fixed effects is more likely to estimate the true relationship than OLS. As prices rose over the period in question, dummies for each year were therefore created to control for this. Prices also showed seasonality, an explanation for which was uncovered in research was that there is a “glut” of animals every august, which depresses prices in the autumn. To control for this seasonality a dummy was created for the months of August, September and October.

To control for variations in the quality of cattle each beef factory receives, control variables were created. These are the percentage of steers with fat scores 3 and higher, and separately the % of animals with conformation of R or better.

The exchange rate of euro with sterling was included to account for variations in demand in the United Kingdom, which is the largest importer of Irish beef produce.

Model

The regression can therefore be shown as

𝒍𝒏𝑷𝒓𝒊𝒄𝒆𝒊𝒕= 𝜶𝒐 + 𝜷𝟏𝑪𝑹𝟑𝒊𝒕+ 𝜷𝟐𝑪𝒐𝒏𝒕𝒓𝒐𝒍𝒔𝒊𝒕+ 𝒖𝒊𝒕

This regression was calculated with CR3 determined by market boundaries from 100km to 300km, rising in increments of 50km.

2.3.5 Entry and Exit Analysis

The effect of a factories entry or exit of a beef factory on cattle prices can provide insight on

competition within the market. If an entrant causes prices to rise, this would indicate that the entrant imposes a competitive constraint on the incumbents, and that the incumbents had previously held some oligopsony power. Competition authorities have used such analyses to determine whether a merger

(18)

18 would be likely to result in a significant lessening of competition on the seller side of markets. In

Airways/American Airlines (2013) the EC used entry and exit to estimate the likely effect of this merger on routes in which both airlines operated, by estimating the effect on prices that the exit of US airways to other similar routes had (Wirth & Akinrinade, 2014). Similarly in The Original Bowling

Company/Bowlplex merger the CMA conducted statistical analysis on the effect of entry or exit of rivals (Wirth & Akinrinade, 2014).

The impact of entry and exit can be examined using a Difference-in-Difference (“DiD”) approach. This technique allows for the estimation of the impact of entry or exit, by comparing the change in the market before and after entry/exit to the change in a control market. The validity of this approach relies on the assumption that the control market shares common trend in demand and cost factors over this period. As this is a strong assumption there must be a strong case for the markets experiencing similar trends.

The market for live cattle for slaughter in Ireland is very suitable for DiD estimation of the effect of entry or exit on prices. Firms compete in location and price and there are no other differentiating factors in a farmers’ choice between factories. Therefore the effect of entry on price can be easily isolated.

The treatment group are the 5 factories closest to the location of the entering or exiting firm, with the 5 factories furthest serving as controls. These groups can be expected to share many of the same cost and demand factors. The control should be less effected by entry/exit, being located further from the entering or exiting firm.

The twelve months prior and post entry were taken in each case. The twelve months prior were the before period, with a dummy signifying the post entry period. A dummy was then created for the treatment factories. From these dummies an interaction term was created which would calculate the DiD.

Therefore in the first EEA regression I test the impact of the entry of Ballon Meats on the nearest five plants, and in the second EEA regression I test the impact of the exit of Liffey Meats.

Both regressions can therefore be shown as

𝒍𝒏𝑷𝒓𝒊𝒄𝒆𝒊𝒕 = 𝜶𝒐 + 𝜷𝟏𝑷𝒐𝒔𝒕𝒕+ 𝜷𝟐𝑵𝒆𝒂𝒓𝒆𝒔𝒕𝒑𝒍𝒂𝒏𝒕𝒊+ 𝜷𝟑(𝑷𝒐𝒔𝒕𝒕 𝒙 𝑵𝒆𝒂𝒓𝒆𝒔𝒕𝒊) + 𝒖𝒊𝒕

Table 2: Summary of variable definitions

𝒍𝒏𝑷𝒓𝒊𝒄𝒆𝒊𝒕 The log of the price of an R3 Steer at slaughterhouse i, in period t.

𝑪𝑹𝟑𝒊𝒕 % of factories owned by the Big Three, within a given radius of slaughterhouse

i, in period t.

𝑭𝒂𝒕𝒊𝒕 % of animals with a higher Fat content than 3 at slaughterhouse i, in period t.

𝑪𝒐𝒏𝒇𝒐𝒓𝒎𝒂𝒕𝒊𝒐𝒏𝒊𝒕 % of animals with a conformation score better than 3 at slaughterhouse i, in

period t.

𝑺𝒆𝒂𝒔𝒐𝒏 dummy variable for the months of August, September and December

𝑬𝒖𝒓𝒐𝒕 euro Sterling exchange rate in period t

𝑷𝒐𝒔𝒕𝒕 dummy variable for the period post entry or exit

(19)

19

4. Results

In this section I describe results of the PCA and EEA regressions.

4.1 Price Concentration Analysis

Table 3: Regressions with and without Fixed Effects

(1)

(2)

(3)

(4)

300km OLS

300km Plant

Fixed

Effects

250km OLS

250km Plant

Fixed Effects

CR3

-0.0006

(-1.29)

-0.0304

***

(-6.29)

-0.0002

(-1.00)

-0.0086

**

(-3.02)

Euro/Sterling

0.0016

*

(2.30)

0.00105

(1.52)

0.0016

*

(2.30)

0.0014

*

(2.01)

Fat

-0.0005

***

(-4.28)

-0.0006

***

(-4.58)

-0.0005

***

(-4.29)

-0.0006

***

(-4.50)

Conformation

-0.0002

**

(-2.69)

-0.0001

(-0.53)

-0.0002

*

(-2.57)

-0.0000

(-0.31)

season

-0.0294

***

(-12.71)

-0.0307

***

(-13.37)

-0.0293

***

(-12.70)

-0.0299

***

(-12.91)

Constant

5.829

***

(88.03)

7.236

***

(30.69)

5.813

***

(91.42)

6.223

***

(40.99)

Adj. R-squared (within)

0.7158

0.7235

0.7156

0.7176

Adj. R-squared (between)

0.3193

0.0003

0.3176

0.0001

Adj. R-squared (overall)

0.7088

0.1707

0.7087

0.3356

Observations

1460

The CR3 coefficient is statistically significant with fixed effects only. As the market size increases, so too do the significance and sign of the coefficient of concentration. The results for regression smaller than 250km are not included here.

The R2 _{for both OLS regressions is high, which shows that a large amount (70%) of the variance in prices I} explained by the independent variables at 300km and 250km. The overall R2 _{for the fixed effects}

regressions is lower at 250km, and even more so at 300km. This is as expected as this analysis does not include the variance between plants, which accounts for a large of the total variance. The loss of variation-between-firms has reduced the R2_{at 250km far less, indicating that this fixed effects}

regression is more reliable. This is sensible, as factories had more variation in the CR3 at 250km, giving this regression more CR3-price relationships with which to estimate the coefficient.

We can reject the null hypothesis that the percentage of plants owned by the Big Three is not correlated with prices at a market definition of 250km and 300km.

(20)

20 The sign is negative, which shows a negative relationship between slaughterhouse ownership and the prices paid to steers, as predicted by the model of oligopsony power.

The coefficient suggests that a one per cent increase in concentration, is linked to a decrease in prices of between 1-3%. The higher end of this range seems implausibly high. To extrapolate this an exogenous ten percent increase in the slaughterhouse ownership of in plant would result in a 8-30% decrease in prices. This seems high given the fact that many independent firms remain in operation, and that they are spread around much of the country, with the exception of the South West.

The 300km FE coefficient predicts that the ABP merger should lead to a nationwide decrease in prices of over 20%, which like the prediction for a 10% increase in the CR3 seems unrealistically high.5_The coefficient for the fixed effects at 250km, predicts a moderate decline in prices of 6.8%, which is more credible.6_{As this estimate is derived from the regression with the greater R}2_{, and which enjoy greater} variation in the CR3 measure, I take this to be the more realistic coefficient.

The increase in the size and significance of the CR3 coefficient is large. This may be evidence of the dampening effects of the omitted variables, such as proximity to supply and ports. There is far more variation between plants than within plants measure of competition. Therefore an endogenity in the competition measure between plants could easily outweigh link that is apparent in the variation between plants.

5_{The ABP merger should increase the CR3 by 8% nationwide, leading to a predicted price change of -24%.} 6_{The ABP merger should increase the CR3 by 8% nationwide, leading to a predicted price change of -24%.}

(21)

21

4.2 Entry and Exit Analysis

Table 4: Entry and Exit Analysis results

(1)

(2)

Liffey Meats (Exit) Ballon Meats (Entry)

Nearestfive

-0.0121

(-1.81)

-0.00464

(-0.68)

DummyPost

0.0547

***

(7.84)

0.0445

***

(6.39)

DiD

0.00484

(0.49)

0.00264

(0.27)

Constant

5.983

***

(1266.26)

6.012

***

(1250.10)

Adj. R-squared

0.3692

0.2630

Observations

240

249

The sign and coefficient of the Post dummy variable is significant and positive as we should expect as prices were increasing across periods in both cases.

The coefficient for Nearest indicates that in both cases the factories nearest the sites of entry/exit had marginally lower prices. However as the coefficient lack significance we cannot say whether or not this is due to random variation. The size of the effect is so small as to have been economically insignificant. The DiD estimator is not statistically significant for either the exit of Liffey Meats, or the entry of Ballon. This shows that entry and exit doesn’t not effect local firms to a greater extent than more distant firms.

(22)

22

5. Discussion

With these results in mind, I now return to the two questions that motivated this research.

Is the level of concentration in factory ownership leading to lower cattle prices?

The negative relationship from the FE model can be hard to square with entry and exit analysis finding that entry had no effect on prices. However in a national market EEA could not be expected to identify the true casual effect. If the market is near national in scope, entry or exit should be expected to impact prices paid by factories across Ireland. This means the factories in the control group would have been affected by entry/exit also. Therefore the EEA approach would at best capture the difference between plants that are affected, and those that are less effected.

The PCA results are what would be expected were entry and exit having an effect on prices across the entirety of Ireland. The negative sign of this effect is in line with theoretical predictions of the

Oligopsony model, as with greater concentration of slaughterhouse ownership prices fall. Furthermore as this effect is only evident with a market definition of 250km-300km it is in line with our evidence of a near-national geographic market from prices. Furthermore the results are strongly statistically

significant. Therefore I accept the finding of these results, that CR3 is negatively correlated with prices, indicating that concentration leads to lower prices. As discussed in the results, the coefficient for CR3 at 250km is likely the more reliable, suggesting that an increase in CR3 of 10% begets a decrease in prices of 8.6%.

The EEA and PCA results therefore point to a market in which one independent slaughterhouse entering or exiting can effect competition not only among nearby factories, but across Ireland due to the “chain of substitution” effect. Does the level of concentration effect prices? Yes, as the addition or removal of an independent firm can impact prices nationwide.

Should the ABP Fayne Valley merger be expected to lead to lower prices for live cattle?

The main area of concern I identified pre-estimation was the Southern half and South Western quarter of Ireland. The merger here creates a stronghold of 6 of 11 factories in Southern Ireland for ABP, and 9 of 11 factories the Big Three. However evidence from both the EEA and factory prices, suggests that any impact on competition will affect prices nationally.

The ABP/Linden merger with an incremental market share of 8%, resulted in a CR3 of almost 70% in live cattle purchasing. As outlined earlier, such concentration in purchasing can create buyer power, which when held in the face of competitive supply can lead to an allocative efficiency loss.

The results for the PCA do indicate that an increase in the CR3 in Irish market for live cattle results in lower prices. The predicted impact of the ABP/Linden merger is a roughly 6.9% decrease in prices. This seems reasonable, given the large incremental share (8%), and the already high CR3 (65%).

Should the merger be expected to decrease prices in the market for live cattle for slaughter? Yes, as the increase in concentration can be expected to enhance the Big Three’s oligopsony power.

(23)

23

Bibliography

AgriAware. (2013). The Irish Beef industry and CAP.

Alvarez, A., Fidalgo, E. G., Sexton, R. J., & Zhang, M. (2000). “Oligopsony Power with Uniform Spatial Pricing: Theory and Application to Milk Processing in Spain. European Review of Agricultural Economics.

Ashenfelter et al. (2004). Econometric Methods in Staples.

Azzam, A., & Anderson, D. (1996). Assessing Competition in Meatpacking: Economic History, Theory and Evidence. U.S. Department of Agriculture, GIPSA-RR.

Bhattacharyyaa, S., & Nain, A. (2011). Horizontal acquisitions and buying power: A product market analysis. Journal of Financial Economics.

Bordbia. (2016). Factsheet on the Irish Agriculture and Food & Drink Sector.

Chen, Z. &. (1992). Supply Analysis in an Oligopsony Model. American Journal of Agricultural Economics, 74(4), 973-979. Retrieved from http://www.jstor.org/stable/1243195

Chen, Z. (2007). Buyer Power: Economic Theory and Antitrust Policy. Research in Law and Economics, pp.17 - 40.

CMA. (2010). Merger Assessment Guidelines.

Crespi , J. M., Sexton, R. J., & Saitone, T. L. (2012). Competition in U.S. Farm Product Markets: Do Long-Run Incentives Trump Short-Long-Run Market Power? Appl Econ Perspect Policy.

Crespi , T. L., Richard, S. J., & Sexton, J. M. (2012). Competition in U.S. Farm Product Markets: Do Long-Run Incentives Trump Short-Long-Run Market Power? Appl Econ Perspect Policy, 34 (4): 669-695. DAFM. (2016). Beef Carcase Classification and Price Reporting Section Annual Report 2016.

Department of Agriculture, F. a. (2009). Irish Meat Market.

Dobson, P. W., Clarke, R., & Davies, S. (2001). Buyer Power and its Impact on Competition in the Food Retail Distribution Sector of the European Union. Journal of Industry, Competition and Trade. European Commission. (2016). Case M.7930 - ABP GROUP / FANE VALLEY GROUP /SLANEY FOODS. Garces and Davis. (2009). Quantitative Techniques for Competition and Antitrust Analysis.

Hackl et. al. (2014). Market Structure and Market Performance in E-Commerce.

Huang, S., & Sexton, R. J. (1996). Measuring Returns to an Innovation in an Imperfectly Competitive Market: Application to Mechanical Harvesting of Processing Tomatoes in Taiwan. American Journal of Agricultural Economics.

Love, A. H., Capps , O., & Williams, G. W. (2009). CONCENTRATION IN BEEF PACKING AND SLAUGHTER CATTLE PRICING. TAMRC Commodity Market Research.

(24)

24 Macconnell, S., (2000, 29 March 2000). No evidence of price-fixing cartel, Bacon claims in report for meat plants. Received from: http://www.irishtimes.com/news/no-evidence-of-price-fixing-cartel-bacon-claims-in-report-for-meat-plants-1.260788

Manzouk and Moul. (2008). Prices and endogenous market structure in office supply superstores. Mazzeo, M. (2002). Product choice and oligopoly market structure.

McCloughan, P. (2016). Submission to the DG regarding proposed transaction in the beef processing market. PMCA Consulting.

Motta, M. (2004). Competition Policy; Theory and Practice.

Pepall et. al,. (2008). Industrial Organization: Contemporary Theory and Empirical Applications.

Raper, K. C., Love, A. H., & Shumway, C. R. (2006). Determining Market Power Exertion Between Buyers and Sellers. Journal of Applied Econometrics.

Rogers, S., (2004, 11 November 2004). IFA: Farmers getting less for beef but cost up in shops. Received from: http://www.irishexaminer.com/ireland/ifa-farmers-getting-less-for-beef-but-cost-up-in-shops-297199.html

Salop, S. C. (1979). Monopolistic competition with an outside good.

Sexton, R. J. (2012). Market power, Misconceptions, and modern agricultural economics. American Journal of Agricultural Economics.

Sexton, R., & Zhang, M. (2001). An Assessment of the Impact of Food IndustryMarket Power on U.S. Consumers. Agribusiness.

Sheldon, I. M. (2017). THE COMPETITIVENESS OF AGRICULTURAL PRODUCT AND INPUT MARKETS: A REVIEW AND SYNTHESIS OF RECENT RESEARCH. Journal of Agricultural and Applied Economics, Sheldon, Ian M., (2017), THE COMPETITIVENESS OF AGRICULTURAL PRODUCT AND INPUT MARKETS: A REVIEW AND SYNTHESIS OF RECENT RESEARCH, Journal of Agricultural and Applied Economics, 49, issue 01, p. 1-44., p. 1-44.

Stiegert, K. M., & Hamilton, S. F. (1998). Backward Implicit Contracts, Pre-Commitment and Market Power in the International Durum Wheat Market. Selected Paper, the American Agricultural Economics Association’s Annual Meeting, Salt Lake City,UT, August 2-5. .

TCA. (2003). MERGER DETERMINATION: The proposed acquisition of Galtee Meats. The Competition Authority. (2003). M/03/029 - Dawn Meats / Galtee Meats. The Office of Fair Trade. (2010). Merger Assessment Guidelines.

Thomas, S., & Fee, C. E. (2004). Sources of gains in horizontal mergers: evidence. Journal of Financial Economics.

Thousands of farmers protest over beef prices (2014, 28 October 2014) Received from:

(25)

25 Ward, C. E. (2002). A Review of Causes for and Consequences of EconomicConcentration in the U.S.

Meatpacking Industry. Current Agriculture, Food & Resource Issues.

Wilcox, M. D., & Abbott, P. C. (2006). Can Cocoa Farmer Organizations Countervail Buyer Market Power? Selected Paper, the American Agricultural Economics Association’s Annual Meeting, Long Beach, CA, July 23-26. .

Wirth, D., & Akinrinade, T. (2014). A Practitioner’s Guide to Price Concentration Analysis. World Competition Law Review.

Xia, T., & Sancewich, B. (2012). Retail Markets and Buyer Power in Agricultural Procurements. Agricultural and Applied Economics Association.

Zheng, X., & Vukina, T. (2009). Do Alternative Marketing Arrangements Increase Pork Packers' Market Power? American Journal of Agricultural Economics.

Is the level of concentration in Irish beef processing lowering cattle prices? : an empirical approach