Quasi-internalization, recombination advantages, and global value chains: Clarifying the role of ownership and control

Abstract

In responding to the Forsgren and Holm (2021) critique of internalization theory, we develop a capability-based model of internalization and quasi-internalization, highlighting the key role of the international recombination of assets. With external control mechanisms becoming more sophisticated, full internalization has become increasingly unnecessary. Rather, the capacity to orchestrate complex networks is an increasingly important source of competitive advantage. We demonstrate that internalization theory does not need to assume that the MNE is all powerful, or that it can dictate the choice of mode with its foreign business partners. We also disagree with the argument that internalization theory presumes perfect rationality. When managers’ perceptions deviate from reality, they do indeed make wrong choices (over- and under-internalization) that come with various types of efficiency penalties. We share the Forsgren and Holm view that a learning perspective can provide insights on the evolution of an MNE’s asset recombination mode, as it gains experience and knowledge. Furthermore, we show that internalization theory has been extended to incorporate such a learning perspective.

Résumé

En réponse à la critique de la théorie de l'internalisation par Forsgren et Holm (2021), nous développons un modèle d'internalisation et de quasi-internalisation basé sur les capacités, en soulignant le rôle clé de la recombinaison internationale des actifs. Les mécanismes de contrôle externes devenant plus sophistiqués, l'internalisation complète est devenue de plus en plus inutile. Au contraire, la capacité à orchestrer des réseaux complexes constitue une source de plus en plus importante de l'avantage concurrentiel. Nous démontrons que la théorie de l'internalisation n'a pas besoin de supposer que l’entreprise multinationale est toute-puissante, ou qu'elle peut dicter le choix du mode de gouvernance avec ses partenaires commerciaux étrangers. Nous sommes également en désaccord avec l'argument selon lequel la théorie de l'internalisation présume une rationalité parfaite. Lorsque les perceptions des managers s'écartent de la réalité, ils font effectivement de mauvais choix (sur- et sous-internalisation) qui s'accompagnent de divers types de pénalités d'efficience. Nous partageons le point de vue de Forsgren et Holm selon lequel la perspective de l'apprentissage peut apporter des renseignements sur l'évolution du mode de recombinaison des actifs d'une entreprise multinationale, à mesure qu'elle acquiert de l'expérience et des connaissances. De plus, nous montrons que la théorie de l'internalisation a été étendue pour incorporer une telle perspective de l'apprentissage.

Resumen

Respondiendo a la crítica de Forsgren y Holm (2021) de la teoría de internalización, desarrollamos un modelo basado en capacidades de internalización y cuasi-internalización, resaltando el papel clave de la recombinación internacional de activos. Al volverse más sofisticados los mecanismos de control externo, la internalización total se ha vuelto cada vez más innecesaria. En su lugar, la capacidad para orquestar redes complejas es una fuente de ventaja competitiva cada vez más importante. Demostramos que la teoría de internalización no necesita asumir que la empresa multinacional es todopoderosa o que puede establecer la elección de modos con sus socios extranjeros de negocio. También discrepamos del argumento que la teoría de internalización supone una racionalidad perfecta. Cuando la percepción de los gerentes se aparta de la realidad, no necesariamente toman elecciones equivocadas (sobre o sub internalización) que viene con varios tipos de sanciones de eficiencia. Compartimos la idea de Forgren y Holm que la perspectiva de aprendizaje puede proporcionar aportes sobre la evolución de modo de recombinación de activos de una empresa mientras va ganando experiencia y conocimiento. Además, mostramos que teoría de internalización ha sido ampliada para incorporar dicha perspectiva de aprendizaje.

Resumo

Em resposta ao comentário de Forsgren e Holm (2021) sobre a teoria da internalização, desenvolvemos um modelo de internalização e quase-internalização baseado em capacidade, destacando o papel fundamental da recombinação internacional de ativos. Com mecanismos de controle externo se tornando mais sofisticados, internalização total tem se tornado cada vez mais desnecessária. Em vez disso, a capacidade de orquestrar redes complexas é uma fonte cada vez mais importante de vantagem competitiva. Demonstramos que a teoria da internalização não precisa supor que a MNE é toda poderosa ou que pode ditar a escolha do modo com seus parceiros de negócios estrangeiros. Também discordamos do argumento de que a teoria da internalização pressupõe racionalidade perfeita. Quando as percepções dos gerentes se desviam da realidade, eles de fato fazem escolhas equivocadas (super e subinternalização) que acompanham com vários tipos de perdas de eficiência. Compartilhamos a visão de Forsgren e Holm de que uma perspectiva de aprendizado pode fornecer insights sobre a evolução do modo de recombinação de ativos de uma MNE, à medida que ganha experiência e conhecimento. Além disso, mostramos que a teoria da internalização foi estendida para incorporar tal perspectiva de aprendizagem.

摘要

针对 Forsgren 和 Holm (2021) 对内部化理论的批评， 我们开发了基于能力的内部化和准内部化模型， 突出了资产国际重组的关键作用。随着外部控制机制变得越来越精致， 全内部化变得越来越没有必要。相反， 协调复杂网络的能力是竞争优势越来越重要的来源。我们提出， 内部化理论不需要假设跨国公司（MNE）是无所不能的， 或者它可以决定与外国商业伙伴的模式选择。我们也不同意内化理论假定完全理性的论点。当管理者的认知偏离现实时， 他们确实会做出错误的选择（内化过度和内化不足）， 从而导致各种效率损失。我们同意 Forsgren 和 Holm 的观点， 即学习视角可提供关于MNE资产重组模式演变的见解， 因为它获得经验和知识。此外， 我们表明， 内化理论已经扩展到包含这一学习视角。

Appendix

Starting from Hennart’s (2009) asset bundling framework, we focus on two firms, an MNE and a local firm in a host market, who have complementary assets that they can bundle in order to generate value in the host market. If they collaborate, they earn a base payoff of ${a}_{0}-{a}_{1}D$, where $D$ is the diversity and distance between the two countries, and each of the subscripted $a$ is a parameter. This is a coalitional payoff, and they bargain over the division of this payoff ex post as described by Nash (1953), splitting the quasi-surplus from collaboration evenly. We assume for simplicity that both firms have outside options of 0.

Suppose that both the MNE and the local firm can make “linking investments” in mutually adapting their assets to each other. This increases the coalitional payoff by the amount $b$. However, the investing firm has to pay an upfront cost of $cD$ in order to make the investment. The resulting payoffs of the two firms, depending on their investment choices, are shown in Figure

Figure 3

Coalition payoff matrix.

3:

This model captures the holdup problem that is described informally by Hennart (2009) and formalized by several scholars, in particular Grossman and Hart (1990). First, we can identify when it is efficient (value maximizing) for the two firms to make the investments. This happens when diversity is below a certain threshold, given by:

$$a_{0} - a_{1} D + 2b - 2cD > 0 \Leftrightarrow D < \left( {a_{0} + 2b} \right)/\left( {a_{1} + 2c} \right) = \overline{D}_{S}$$

while the surplus with investments will be positive as long as

Second, we can identify when the firms actually will make the investments, given that they expect a bargaining process where the holdup problem potentially makes it difficult for them to recoup the investment cost. Since the other firm will be able to appropriate half of the benefit (the addition to the coalitional payoff due to the linking investments), it will be rational to invest only if diversity is below another threshold, given by:

$$b/2 > cD \Leftrightarrow D < b/2c \equiv \overline{D}_{N} = \overline{D}_{E} /2$$

Hence, as long as ${\overline{D} }_{N}<D<{\overline{D} }_{E}$, we get the holdup problem for moderate levels of diversity, reflected in a Prisoner’s Dilemma where neither firm invests, even though they would both be better off if they did.

Finally, note that the level of diversity also determines whether the payoffs, with, as well as without, linking investments, are positive and therefore whether the combination of the assets is feasible. The base surplus (the coalitional payoff without investments) will be larger than 0 as long as

$$a_{0} - a_{1} D + 2b - 2cD > 0 \Leftrightarrow D < \left( {a_{0} + 2b} \right)/\left( {a_{1} + 2c} \right) = \overline{D}_{S}$$

while the surplus with investments will be positive as long as

$$a_{0} - a_{1} D + 2b - 2cD > 0 \Leftrightarrow D < \left( {a_{0} + 2b} \right)/\left( {a_{1} + 2c} \right) = \overline{D}_{S}$$

There are two possibilities: (1) If the investments have a sufficiently attractive cost-benefit ratio ($b/c>{a}_{0}/{a}_{1}$), we get ${\overline{D} }_{S}>{\overline{D} }_{B}$ and so there is a range of diversity in which the asset bundling only takes place if the investments can be guaranteed, and otherwise no entry occurs. On the other hand, (2) if this inequality is not fulfilled, there will be a range of diversity where the bundling will take place but where the linking investments are prohibitively costly. This scenario, which is the one shown in Figure

Figure 4

Influences of O_TI and O_TE on optimal choice of governance mode under different levels of international diversity.

4 below (called “market with no adaptation”), could, for example, respond to a scenario where the MNE pursues a global strategy and exports a standardized product to a local distributor.

Now, we derive the conditions for quasi-internalization, which we model as a self-enforcing contract between independent firms using trigger strategies. Suppose that ${\overline{D} }_{N}<D<{\overline{D} }_{E}$ such that we get a Prisoner’s Dilemma game in Figure 3, in which the suboptimal Nash equilibrium is for both firms to not invest in a one-shot game. In a repeated game, however, each firm can follow the strategy: invest in the first round, and in each subsequent round, as long as everyone has always invested, and do not invest otherwise. What does it take for such a strategy to be a best response to a similar strategy followed by the other firm? Each firm can reap the defection payoff ($\text{(}{a}_{0}+b-{a}_{1}D)/2\text{)}$ by not investing in the first round, i.e., free-riding on the other firm’s investment. However, this triggers the other firm’s retaliation, and hence will ensure the Nash payoff of (${a}_{0}-{a}_{1}D)/2$ forever after. With a discount factor of $\delta$, this gives a net present value of $({a}_{0}+b-{a}_{1}D)/2+({a}_{0}-{a}_{1}D)\delta /2(1-\delta )$. In comparison, cooperating by investing in every round gives a net present value of $({a}_{0}+2b-{a}_{1}D)/2-cD+(({a}_{0}+2b-{a}_{1}D)/2-cD)\delta /(1-\delta )$. It is better to collaborate if the firms are sufficiently forward looking, i.e.,

$$\delta > 2cD/b - 1 \equiv \overline{\delta }$$

This threshold is increasing in diversity (rising from 0 at $D={\overline{D} }_{N}$ to 1 at $D={\overline{D} }_{E}$), suggesting that it is more difficult to uphold a self-enforcing contract with a distant partner where the costs of the linking investments, and thereby the free-riding incentives are high. We assume that $\delta^{\prime}\left( {O_{TE} } \right) > 0$, with the figure shown simply for $\delta ={O}_{TE}$.

Within the Prisoner’s Dilemma range, if $\delta$ is not sufficiently high, the only way to ensure the linking investments is to internalize. To simplify, we do not distinguish here between forward and backward integration, but simply assume that internalization will enable the firms, by means of hierarchical fiat, to ensure that the linking investments are made. Suppose that internalization comes at a hierarchical cost of $h$. In that case, internalization results in a surplus of ${a}_{0}-{a}_{1}D+2b-2cD-h$ while the market (resulting in non-adaptation) results in ${a}_{0}-{a}_{1}D$. The former is higher than the latter if

$$D < \left( {2b - h} \right)/2c \equiv \overline{D}_{I}$$

As long as $h<b$, this threshold is within the range, and hence there is a range ${\overline{D} }_{N}<D<{\overline{D} }_{I}$ where internalization is the optimal solution. Below this threshold, internalization is not necessary, and, above the threshold, the problem that it solves is no longer worth solving. On the other hand, if $h>b$, internalization is prohibitively costly and will never happen. We assume that ${h}^{^{\prime}}\left({O}_{TI}\right)<0$. This implies, as indicated in the figure, that the threshold ${\overline{D} }_{I}$ itself (marked by the vertical black line in the figure) is a function of ${O}_{TI}$: If ${O}_{TI}$ is high, the threshold shifts to the right and then internalization becomes dominant in the space under the diagonal line. On the other hand, if ${O}_{TI}$ is low, the threshold shifts to the left, and it is more likely to get adaptive failure in that space. The figure itself is shown for the following parameter values: ${a}_{0}=4.5$, ${a}_{1}=0.3$, $b=4$, $c=0.4$, and $h=2$.

The stars in the figures show the values for which the results in the matrices in Figure 1 of the paper obtain. The leftmost star is the front panel (low diversity) and the rightmost star is the back panel (high diversity). The middle stars correspond to the middle panel for high and low ${O}_{TE}$, respectively. For high ${O}_{TE}$, we always get quasi-internalization, as indicated by the position of the top star. For low ${O}_{TE}$, on the other hand, it depends on the value of ${O}_{TI}$. When ${O}_{TI}$ is low, ${\overline{D} }_{I}$ shifts to the left and we get market with adaptive failure, whereas with high ${O}_{TI}$ it shifts to the right and we get internalization.

In responding to the Forsgren and Holm's (2021) critique of internalization theory, we have developed a capability-based model of internalization and quasi-internalization, highlighting the key role of the international recombination of assets. With external control mechanisms becoming more sophisticated, full internalization has become increasingly unnecessary. Rather, the capacity to orchestrate complex networks is an increasingly important source of competitive advantage. We demonstrate that internalization theory does not need to assume that the MNE is all powerful or that it can dictate the choice of mode with its foreign business partners. We also disagree with the argument that internalization theory presumes perfect rationality. When managers’ perceptions deviate from reality, they do indeed make wrong choices (over- and under-internalization) that come with various types of efficiency penalties. We share the Forsgren and Holm view that a learning perspective can provide insights into the evolution of an MNE’s asset recombination mode, as it gains experience and knowledge. Furthermore, we show that internalization theory has been extended to incorporate such a learning perspective.