Anti-Poaching Agreements in Labor Markets Oz Shy
Rune Stenbacka
Seminar Presentation Liverpool Business School Liverpool, Sept 29, 2017
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
1 / 27
Introduction
Background and motivation
Background and motivation Employees are the most important asset of a firm. Typically very costly for employers (firms) if experienced productive workers switch to rivals. R&D intensive firms compete aggressively to recruit experienced high-skilled workers. Examples: software engineers, biotechnology, chip designers. Commonly-used recruitment method: Attractive “poaching” offers to workers employed by competitors. Higher salary to induce a worker to switch employers. Firms may have an incentive to form an anti-poaching agreement (commonly referred to as a “no cold call” agreement).
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
2 / 27
Introduction
A Silicon Valley Court case
A Silicon Valley Court case In 2010, the US DOJ’s Antitrust Division filed a complaint against Adobe, Apple, Google, Intel, Intuit, and Pixar. The charge: Bilateral anti-poaching agreements violate Section 1 of the Sherman Act. The argument: Anti-poaching agreements are anti-competitive as they eliminate a significant form of competition to attract high-tech employees, thereby making employees worse off. Subsequent litigation: The antitrust action was followed by a 2013 civil class action, involving 64,446 class members. Settlement: On 9/2/2015, District Court in San Jose approved a $415 million settlement ($5,770 avg. per worker).
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
3 / 27
Introduction
Other cases
Other past and recent cases Anti-enticement laws in Southern US states during 1875-1930 imposed financial fines on planters trying to poach workers with an existing employment contract. Naidu (2010) empirically found that these laws reduced competition and reduced labor mobility. In 1946, Henry Ford II hired many executives from GM after losing a significant market share to GM (more innovative firm). Ciapanna (Rand J., (2011)) analyzes client firms that poach consultants from their consulting company. At AOL Time Warner, 12 out of 19 in the strategy group were hired from McKinsey, Bain, and BCG. American Express hired > 50 from McKinsey in 1 year. WSJ May 31, 2015: Carnegie Mellon University is scrambling to recover after Uber poached 40 of its researchers and scientists, a raid that left a world’s top robotics research institutions in a crisis. Sept. 9, 2016, a former LG sales manager sues LG and Samsung (Northern Calif. Fed. Court) for agreeing not to poach each other’s US employees (LG denied such an agreement ever existed). O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
4 / 27
Introduction
DOJ and FTC antitrust guidance
DOJ and FTC antitrust guidance for human resource professionals October 2016: Naked wage fixing and anti-poaching agreements are per se illegal. ”If the agreement is separate from or not reasonably necessary to a larger legitimate collaboration between employers, it is deemed illegal without any inquiry into its competitive effects”. A per se illegal status seems to require, as a robust feature, that anti-poaching agreements damage market performance. Otherwise, a rule of reason approach would appear to make sense. A rule of reason approach seems consistent with established practice applied to certain other types of agreements, for example RJVs. Our study: What are the pros and cons of a per se illegality policy?
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
5 / 27
Introduction
Research questions
Anti-poaching agreements: Research questions
What are the effects of anti-poaching agreements on profits, worker surplus, and total welfare? What are the main effects of competition for productive workers in the absence of anti-poaching agreements? Note: An anti-poaching agreement is not a mechanism for the coordination of wage setting, because workers always have the option to switch employers if they find it beneficial.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
6 / 27
Introduction
Modeling labor markets
Modeling imperfect competition in labor markets Labor market theory relies heavily on perfect competition (supply & demand curves, firms take market wage as given). Alan Manning (Monopsony in Motion, 2000): 1. Employers are wage setters (not wage takers, as in perfect competition). 2. A wage cut of 1c/ does not cause all workers to leave an employment relationship (as implied by perfect competition). 3. There are important frictions in labor markets (such as heterogeneous preferences, costs of labor mobility). 4. Labor economics should adopt a similar approach to that in industrial organization, and start the analysis from the position that employers have some market power in the labor market.
We apply the imperfect competition approach: heterogeneous switching costs for workers switching and wage-setting firms that can adopt poaching offers in order to implement wage discrimination. O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
7 / 27
Introduction
Modeling method
Modeling method
Duopoly labor market with firms competing for workers. Costly for workers to switch employers. Differentiated switching costs. The productivity of a worker may increase or decrease as a result of a switch to a new employer. “Poaching offers” in labor markets resemble “history-based” pricing in the product markets [examples: Chen (1997), Fudenberg and Tirole (2000), Gehrig, Shy, and Stenbacka (2010), Esteves (2010), and surveys in Villas-Boas (2007) and Esteves (2009)].
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
8 / 27
Introduction
Main Results
Main Results Labor market competition in the absence of anti-poaching agreements 1. Firms pay higher wages to workers who switch from a competing firm. Sliwka and Kampk¨otter (2014): evidence from the German banking industry. 2. In equilibrium, some workers always switch employers [even if switching reduces the productivity of a switching worker]. 3. History-based (employment-based) wage discrimination reduces the intensity of labor market competition. [Opposite result of history-based price discrimination in the product market]. 4. Higher and more differentiated worker switching costs soften labor market competition. [Same result as history-based price discrimination in the product market]
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
9 / 27
Introduction
Main Results (con’d)
Main Results (con’d)
5. Aggregate industry profit increases under anti-poaching agreements (but, one firm may lose under certain conditions). 6. Workers are unambiguously worse off under anti-poaching agreements. 7. Hence, the effects for total welfare of an anti-poaching agreement always reflect a tradeoff between industry profit and worker surplus. 8. With sufficiently strong productivity-enhancing switching, anti-poaching agreements always reduce total welfare. 9. With moderately productivity-enhancing switching combined with high switching costs or with productivity-reducing switching, anti-poaching agreements improve total welfare.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
10 / 27
The model
The labor market
The model: The labor market
Two firms (employers) labeled A and B. wA and wB are wages paid to “loyal” employees. vA and vB are wages paid to “poached” employees. Under an anti-poaching agreement, wA = vA and wB = vB . 2n workers: Initially, n are employed by A and n by B.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
11 / 27
The model
Workers and productivity
The model: Workers and productivity A worker produces k units of output in the present employment relationship. A worker produces k 0 units of output when switching to a rival firm. Definition: We say that switching between employers is (a) productivity-enhancing if it increases the worker’s productivity (k 0 > k); (b) productivity-reducing if it decreases the worker’s productivity (k 0 < k); (c) productivity-neutral if the worker’s productivity does not change (k 0 = k).
k 0 − k measures the magnitude of productivity change associated with switching.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
12 / 27
The model
Labor mobility and switching costs
The model: Labor mobility and switching costs Workers differ in their cost of switching employers Cost of switching is σs, where σ > 0 and s ∈ [0, 1] [worker-specific] The utility function of a worker s who initially works for firm i is defined by ( wi if continues to work for firm i Ui (s) = vj − σs switches to work for firm j, Workers with high switching costs (high s) stay loyal Workers with low switching costs (low s) switch employers 0
n si workers who switch to from i to j
Switching thresholds are: sA = O. Shy, R. Stenbacka
vB − wA σ
Anti-Poaching Agreements
si stay loyal to i
and sB =
-s
1
vA − wB σ
September 29, 2017
13 / 27
The model
Cost, output, and profits
The model: Cost, output, and profits Cost (total wage bill) of firm i: ci (wi , wj , vi , vj ) =
n(1 − si )wi | {z }
+
loyal workers’ wages
nsj vi |{z}
poached workers’ wages
Output of firm i: qi (wi , wj , vi , vj ) = n(1 − si )k + | {z } loyal workers
nsj k 0 | {z }
workers switching from j
Firm i chooses loyalty and poaching wages to maximize expected profit: max πi (wi , wj , vi , vj ) = pqi − ci , wi ,vi
where p (output price) is exogenously given. O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
14 / 27
Equilibrium with poaching
Wage analysis
Poaching equilibrium: Analysis of equilibrium wages
wA = wB =
p(2k + k 0 ) − 2σ 3
and vA = vB =
p(k + 2k 0 ) − σ . 3
(a) All wages decline with switching costs: wi ↓, vi ↓ when σ ↑ (b) There is a wage premium for switching workers: vA > wA and vB > wB (premium increases with switching cost parameter, σ). (c) All wages increase with output price and the productivity parameters: wi ↑, vi ↑ when p ↑, k ↑, k 0 ↑.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
15 / 27
Equilibrium with poaching
Labor mobility and switching
Poaching equilibrium: Labor mobility and switching
The fraction of switching workers: sA = sB = 0
n si workers who switch from i to j
1 3
+
p(k 0 −k) 3σ
si stay loyal to i
-s
1
si ≥ 1/3 ⇐⇒ k 0 ≥ k (switching enhances productivity) si = 1/3 if k 0 = k (switching has no effect on workers’ productivity) Some workers switch employers even when switching is unproductive (k 0 < k)
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
16 / 27
Equilibrium with poaching
Equilibrium profits
Equilibrium profits Industry profit: Π = πA + πB =
2n [p(k 0 − k) − σ]2 + p 2 (k 0 − k)2 + 4σ 2 . 9σ
Equilibrium industry profit increases with the switching cost parameter (σ). Industry profit increases with the output price (p). Industry profit increases with the productivity change induced by switching (k 0 − k) if and only if σ ≤ 2p(k 0 − k). Note: If the last condition is reversed, poaching competition becomes too costly for firms.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
17 / 27
Equilibrium with poaching
Worker surplus and total welfare: Definitions
Worker surplus (WS) and total welfare (W): Definitions Total labor income equals firms’ total wage cost: I = cA + cB . Worker surplus equals total income (wages) minus switching costs: WS = I − SC , where 0
n si workers who switch from i to j Z SC = n
sA
Z σs ds + n
0
sB
σs ds = 0
si stay loyal to i
-s
1
n [p(k 0 − k) + σ]2 , 9σ
Total welfare equals worker surplus plus industry profit: W = WS + Π, where Π = πA + πB
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
18 / 27
Equilibrium with poaching
Profits, worker surplus, total welfare: Graphs
Profits, worker surplus, total welfare: Graphs
Total welfare: W = Π + WS
← more (labor competition) less →
W Π WS
σ max
σ min
σ
Worker surplus and total welfare decrease when switching becomes more costly. Profits increase (relaxed competition). O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
19 / 27
Anti-poaching agreements
Anti-poaching agreements: Wages
Anti-poaching agreements: Wages Definition: Competing firms (employers) form an anti-poaching agreement if they agree not to apply poaching wages in order to attract workers from the rivals. Formally, firms are restricted to setting uniform wages, denoted by uA = wA = vA and uB = wB = vB . Workers have no incentives to switch employers with identical wages (uA = uB ) , hence we analyze an asymmetric equilibrium where uA ≥ uB . Workers switch only from B to A, so that the fractions of switching 0 workers sAAP = 0 and sBAP = p(k3σ−k) , for k 0 ≥ k. Note: ‘Interesting’ equilibrium because symmetric firms end up in an asymmetric equilibrium!
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
20 / 27
Anti-poaching agreements
Anti-poaching agreements: Wages
Effects of anti-poaching agreements on wages Suppose k 0 > k. Then, an anti-poaching agreement reduces the wages paid to all types of workers. Formally, (uAAP < wA < vA and uBAP < wB < vB ); reduces labor mobility not only by eliminating switching from the high-wage firm as sAAP = 0, but also by reducing switching in the other direction as sBAP = sB − 13 . Therefore, an anti-poaching agreement weakens wage competition in the labor market. Note: If k 0 = k, firms pay identical wages uAAP = uBAP = pk − σ and there is no switching.
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
21 / 27
Anti-poaching agreements
Anti-poaching agreements: Profits
Effects of anti-poaching agreements on profits
Industry profit
Suppose k 0 ≥ k. Then, an anti-poaching agreement increases the profit of the high-wage firm. The high-wage firm attracts some workers to switch from the rival firm and it earns a higher profit than the competing firm. Formally, πAAP − πBAP = np(k 0 − k)/3 ≥ 0; the profit of the high-wage firm relative to the poaching equilibrium. Formally, πAAP ≥ πA ; the profit of the low-wage firm relative to the poaching equilibrium if and only if workers face sufficiently high switching costs. ← more (labor competition) less → Formally, πBAP ≥ πB if and only if √ def σ ≥ σBAP = 1 + 2 p(k 0 − k)/2. Π aggregate industry profit relative Π to the poaching equilibrium. =⇒ Formally, ΠAP > Π. AP
σ max
σ min
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
σ
22 / 27
Anti-poaching agreements
Anti-poaching agreements: Worker surplus
Effects of anti-poaching agreements on worker surplus Aggregate worker surplus is lower under the anti-poaching agreement compared with the poaching equilibrium. Formally, WS AP < WS.
Worker surplus
← more (labor competition) less →
WS AP WS
σ max
σ min O. Shy, R. Stenbacka
Anti-Poaching Agreements
σ
September 29, 2017
23 / 27
Anti-poaching agreements
Anti-poaching agreements: Total welfare
Effects of anti-poaching agreements on total welfare √
Define:
26 def 2 + 2 + z = √ 26 2+ 2 −
2 3 1 3
√ >1
and
AP
σW
def =
2+
26
2
!
0
p(k − k).
(a) If k 0 > zk, an anti-poaching agreement always reduces total welfare compared with the poaching equilibrium. (b) If k ≤ k 0 < zk, an anti-poaching agreement increases total welfare AP . compared with the poaching equilibrium if and only if σ ≥ σW ← more (labor competition) less →
W AP W
σ min O. Shy, R. Stenbacka
W AP W
Total welfare (k 0 < zk)
Total welfare (k 0 > zk)
← more (labor competition) less →
σ max
σ
σ min
Anti-Poaching Agreements
AP σW
σ max September 29, 2017
σ
24 / 27
Conclusions
Conclusions: Summary of results Endogenous variable
Effects of anti-poaching agreements
Fraction of switching workers (si )
Reduced
Aggregate worker surplus (WS)
Reduced
Aggregate industry profit (Π)
Enhanced (small firm may lose) Reduced if (k 0 > zk)
Total welfare (W )
AP ) or (k ≤ k 0 < zk and σ < σW
Enhanced if (k ≤ k 0 < zk AP ) or (k 0 < k) and σ > σW
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
25 / 27
Conclusions
Potential Extensions
(a) Separating the expected magnitude from the dispersion of the switching costs. [Characterization in the appendix] (b) Labor force heterogeneous not only with respect to switching costs, both also w.r.t. productivity. (c) Interaction between strategic competition in both the labor and product markets. [Anti-poaching agreements may have different effects on industry profit and welfare in the presence of strategic product market considerations.]
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
26 / 27
Conclusions
Conclusions
(a) Access to industry-specific facts are needed in order to verify whether an anti-poaching agreement could prevent firms from engaging in excessively aggressive recruitment of human capital from rivals. (b) Even though one should not exaggerate the number of industries where anti-poaching agreements could be beneficial, our analysis justifies a policy whereby anti-poaching agreements should be judged on a case by case basis, giving firms an option to present an efficiency defense. [In line with the intuitive comment made by Heyer and Shapiro (2010).]
O. Shy, R. Stenbacka
Anti-Poaching Agreements
September 29, 2017
27 / 27