Behavioural influences in Portuguese foreign direct investment

The Journal of Socio-Economics 40 (2011) 394–403

Contents lists available at ScienceDirect

The Journal of Socio-Economics journal homepage: www.elsevier.com/locate/soceco

Behavioural influences in Portuguese foreign direct investment Ricardo Pinheiro-Alves a,b,∗ a b

Universidade da Beira Interior, Dep. Gestão e Economia, Estrada do Sineiro, 6200-209 Covilhã, Portugal IADE, Av. D. Carlos I, 4, 1200-649 Lisboa, Portugal

a r t i c l e

i n f o

Article history: Received 11 September 2009 Received in revised form 6 September 2010 Accepted 5 October 2010 JEL classification: D21 F21

a b s t r a c t The paper presents a behavioural economics approach to FDI. It relies on questionnaires and interviews with Portuguese managers to present evidence of the role played by herding, anchoring, mental accounting and other behavioural rules in FDI location decisions. It originates a set of heuristics influencing the direction of FDI flows and it confirms the prediction of the Heiner model (1983, 1989) that the higher the uncertainty faced by decision makers the more frequent will be the use of behavioural rules. The results go beyond neoclassical theory by helping to explain non-maximizing decision-making by managers. © 2010 Elsevier Inc. All rights reserved.

Keywords: Behavioural economics Uncertainty FDI determinants Decision making

1. Introduction The behavioural finance literature has shown that investment decisions in equity markets cannot be totally explained by a neoclassical approach (e.g. Shiller, 2003). FDI theory seldom considers the role of managers within an investment decision making process. However, psychologists recognize that managers have several motivational factors that are either intrinsic to their personality or shaped by their environment, and that their choices change with individual personal experience (Frey and Eichenberger, 2001). Although managers have checks on their performance (from competition, shareholders, customers and employees) and thus they often do make their choices more carefully than as if they acted as individuals, they are not immune to moral, cultural and other social influences usually disregarded by economic literature. Hence it is important to understand the different perceptions of managers and their effects on real life FDI location decisions. The aim of this paper is to enrich the contribution of the behavioural approach to FDI theory by apprehending the economic significance of heuristics in manager’s investment decisions. These heuristics are rules of behaviour repeatedly followed by managers that influence firms in the choice of external markets. The approach

∗ Correspondence address: Gabinete de Estratégia e Estudos, Av. da República, 79, 1◦ , 1050-243 Lisboa, Portugal. Tel.: +351 919607069/217998157; fax: +351 217998154. E-mail addresses: [email protected], [email protected] 1053-5357/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.socec.2010.10.002

followed in the paper gives a central role to the uncertainty faced by managers (the Heiner model) in explaining the existence of behavioural rules. The following section briefly reviews FDI theory while Section 3 details the applied methodology. Section 4 presents empirical evidence of behavioural rules and Section 5 deals with the role of uncertainty by testing the Heiner model. The paper concludes with the consequences of the use of behavioural rules in FDI decision making. 2. A behavioural approach to FDI Consider a firm deciding whether to invest abroad and where to locate its investment. An economically rational decision-maker attempts to maximize the net present value of the difference between revenue and costs when answering these questions. These are the two key variables for a decision and economic literature presents several explanations based on potential revenue and costs for FDI to occur. Multinational enterprises (MNE), when making FDI location decisions within imperfect markets, seek to improve their revenue stream in several ways. They use specific advantages (such as product differentiation, managerial and marketing skills, innovation and technology or scale and agglomeration economies) over local competitors in the host market to compensate the additional costs of investing abroad. The will to minimize transactional costs and thus to be more cost-efficient is also used by the FDI literature to explain location decisions (Williamson, 1981). Buckley and

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

Casson, Hennart, Caves and others further developed this approach by stating that the resulting power of market imperfections (originating in less-tradable goods such as “research and development”, knowledge or intangible assets such as brands) are an incentive for internalization and thus for the formation of MNE (Caves, 1996). Further explanations of location decisions are mainly related with the fragmentation of production processes by single-plant firms into different stages based on different relative factor endowments, and thus prices, across countries. In this case, vertical FDI tends to be unidirectional, from richly endowed countries to cheaper labour endowed locations (Blonigen, 2005). The proximityconcentration model (Horstmann and Markusen, 1992; Brainard, 1993) explains multi-plant MNE and two-way horizontal FDI when it becomes relatively less expensive in comparison with exporting. The above explanations usually assume firms and managers as rational profit maximizers where uncertainty is often reduced to risk so that rationalization conditions can be developed. The traditional behavioural approach to FDI, on the other hand, states that the relevant factor for the location decision is psychic distance, that is, “. . . the sum of factors preventing the flow of information from and to the market. Examples are differences in language, education, business practices, culture and industrial development” (Johanson and Vahlne, 1977, p. 24). In this approach, FDI location decisions require a huge amount of information and the decision making process involves a lot of different people that, directly or indirectly, influence the final location. Moreover, they comprise different steps where a large number of small sequential decisions are made during several months before the final decision to immobilize capital for several years is reached (Aharoni, 1999). While the decision making process evolves, environmental variables are permanently changing in unpredictable ways and decision makers are themselves affected by rather different events, thus giving uncertainty a central role. Uncertainty is here seen differently from expected utility theory where it is reduced to risk acting as a constraint to maximization. Uncertainty is “the absence of ability to decipher all of the complexity of the environment; especially one whose very structure itself evolves over time” (Heiner, 1983, p. 569). It includes, besides risk, “nonreplicable uncertainty or even ignorance” (Heijdra, 1988, p. 83) present in most situations faced by decision makers. These cannot be mitigated and prevent the assignment of probabilities for each alternative in the decision making process (Knight, 1921). In the behavioural approach, each FDI location decision comprises not only the “economically rational” part but also the “behavioural” part, where perceptions and other cognitive features of managers, in an uncertain context, are included (Katona, 1975). Therefore, a behavioural approach should consider how the extrinsic and intrinsic cognitive characteristics of managers, the basis of their changing expectations, influence each step of the decisionmaking process. In the presence of uncertainty, managers tend to rely on behavioural rules or heuristics, that is, simplifying strategies to reduce complexity that systematically deviate from the predictions of unbounded procedural rationality (Frey and Eichenberger, 2001). Managers are thus subject to errors and “anomalous” behaviour in decision making. Both may be corrected. But while errors are single deviations from economic rationality explained by the limited capabilities of human beings, heuristics are sequential deviations represented by systematic and predictable biases arising from behavioural rules. In a dynamic perspective, when managers are finally able to correct their anomalous behaviour the environment has changed in a significant way and, because a changing context modifies their perceptions, they have to permanently re-start their personal learning process to cope with the new environmental conditions. All heuristics that are recurrent and persist during a certain period of time because they are not immediately corrected through learning or incentives may be considered as behavioural

395

Table 1 Taxonomy of behavioural rules in FDI decisions. Type time

Intrinsic

Extrinsic

Past

Learning, hindsight bias, sunk costs, mental accounting, break-even effect, house money effect Framing, representativeness

Historical anchoring, cultural anchoring

Present

Future

Strategic inconsistencies Overconfidence, confirmatory bias

Availability, feelings, fairness, herding, cascading, signalling, false consensus bias, reputation-based herding, inter-expert inconsistency Strategic inconsistencies

rules (Arrow, 1996). The Heiner (1983, 1985, 1989) model, theoretically applied to FDI in Hosseini (2005), emphasizes this “rigidity” in decision-making by underlining the objectives and motivations of any individual, e.g. managers investing abroad, in a way that allows the identification of relevant durable patterns in his behaviour. The behavioural literature has shown that rules of behaviour do exist in investment operations or similar situations: Mental accounting (Thaler and Johnson, 1990); Stategic inconsistency in firms’ decisions (Schwartz, 1998); Overconfidence (Hilton, 2003; Malmendier and Tate, 2005) and confirmatory bias (Rabin, 1998); Anchoring (Frey and Eichenberger, 2001; Grinblatt and Keloharju, 2001; Beckmann et al., 2008); Availability (Tversky and Kahenman, 1982); Herding (Banerjee, 1992; Zwiebel, 1995; Kinoshita and Mody, 2001); Fairness (Kahneman et al., 1986). Many of these heuristics have been found in financial markets and, although the actions and the outcomes of these markets are much more easily observable than in the case of foreign investment, some may be extrapolated to explain information collection, selection of alternatives and the final FDI location decision. Table 1 presents the taxonomy of behavioural rules typified in accordance with their time reference, that is, related with past or present events or concerning expectations about future developments (rows), and by its source of motivation, the intrinsic or extrinsic dimension of cognitive characteristics (columns). This is not an exhaustive list of all behavioural rules but of those that could apply to FDI operations. Given the large number of heuristics the paper is focused only on a subset. 3. Methodology 3.1. Data In order to find evidence of behavioural rules in FDI location decisions, a database was built from information collection, and interviews and questionnaires applied to managers of the largest 50 Portuguese firms with FDI. A total of 112 operations abroad were considered representing at least 6% of the total FDI (Banco de Portugal, 2005).1 These operations cover a range of industries: agriculture, manufacturing, energy, construction, financial and services. The common denominator is that all operations represent part of a firm’ production capabilities installed abroad. The reason is that the decision to invest abroad has to be very well thought and the uncertainty associated significantly greater than, for example, the opening of a representative office (usually a support for exports). Managers included in the study were directly responsible

1 The value is higher but Banco de Portugal did not provide information on the number of FDI locations without a productive component, such as representative offices.

396

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

for, or participated in, 76% of the total number of FDI decisions here considered and the actual management of the firms is the same, or follows a similar internationalization strategy, in 88% of operations. In Portugal, most of the largest firms are either influenced by the state or family owned businesses. Each operation corresponds to a country location by a Portuguese investor. The sample is skewed for large firms in Portugal although these are medium size firms in international terms. Both the questionnaires and the interviews dealt with FDI operations, namely how they were incorporated within the strategy of each firm and their respective decision-making procedures. Interviews were implemented using a set of questions covering the FDI decision-making process of each firm and having in mind the heuristics referred in the literature. The approached themes served as a grid for the content analysis of the information collected (Neuendorf, 2002). First, a case description for each FDI operation was organized by the pre-determined themes. A coding of the information was defined following the pre-set criteria from the research hypothesis that FDI location decisions are influenced by rules of behaviour and applied to each theme. A cross-case analysis was then performed based on quantitative data obtained from the coding and leading to the results for each type of heuristic. Data collection also relied on supplied documentation and other available information both in firms’ internet sites, such as annual reports, or on other sources. Documentation analysis was used both complimentary and as a source of validity for some of the collected information. The identification of behavioural rules was based both on the procedures followed by each firm in the entire decision-making process and the management of each FDI after their implementation. An example of the first situation happens when FDI data shows the simultaneous choice by many Portuguese firms of a determined location. Then, managers are asked about the motivations to choose that location. If a manager justifies it both with the expected profitability and the existence of a “national” will or a fashion to invest in that market then a situation of “herding” behaviour is considered as an influence to the location decision. An example of the second situation is the decision to finance an investment abroad with domestic earnings obtained from a monopoly situation. This “mental accounting” is confirmed when a firm keeps a loss-making investment operating for several years, which is explained by the manager’s expectation to recover these losses. 3.2. The Heiner model The database also serves to test the Heiner model (1983, 1985 and 1989). The model confronts the “competence” of an agent with the “difficulty” in selecting most preferred alternatives in a decision-making process. Existing gaps between competence and difficulty arise due to uncertainty. Therefore, uncertainty exposes the limits of any agent in any selection process and is the origin of behavioural regularities. Greater uncertainty leads managers to simplify the decision-making process and to augment their reliance on behavioural rules. In a FDI operation, each new decision deals with potential sources of information on costs, revenue and risk. But the access to these sources has a component of uncertainty. Due to uncertainty managers do not know if the selection of new information improves their performance. Their response to potential information depends both on the environment and individual perceptions. These are the two types of variables presented by the model. The first represents the environmental (complexity-stability) influences from the past, present and future, surrounding the decisions made by managers, while the second refers to how managers perceive, through their intrinsic and extrinsic characteristics, the connection between their behaviour and the environment, that is,

Table 2 Identified rules of behaviour.

Intrinsic Extrinsic

Past

Present

Future

30 43

13 86

3 –

how they react to new information. The two together determine the degree of uncertainty (U). The more complex is the environment or the less reliable are the perceptions of managers the greater is the uncertainty, and thus the use of behavioural rules, in the decision making process. Formally, as long as C < D ⇒ U = Ae + Ai > 0 where C and D refer to Competence and Difficulty, and A is divided into the extrinsic (Ae) and intrinsic (Ai) components of behavioural rules. This is valid for all decisions within the FDI process. The central hypothesis of the Heiner model states that the higher the uncertainty the higher is the reliance of decision makers on behavioural rules of intrinsic (Ai) and extrinsic (Ae) nature. A prediction of the model is that firms invest more where there is less uncertainty.2 If this is true then the relevance of uncertainty in decision-making and the usefulness of a behavioural approach to FDI theory is reinforced. The size of the sample and the nature of the data (nominal and rank variables) require the application of non-parametric tests (Norusis, 2003).3 The independence and association tests are performed in the SPSS software, version 12, in order to reveal the strength and the direction of the relationship between uncertainty and the use of behavioural rules. 4. Evidence of behavioural rules The collected information allowed the identification of 175 situations where behavioural rules followed by managers have influenced investment decisions abroad. Table 2 divides them as per the taxonomy presented above. This presentation uses the time span as reference. 4.1. Past events Past events or experiences may influence managers at least during several years. Portugal Telecom (PT) enjoyed, for a long period, a comfortable position as a monopolist provider of telecommunication services in Portugal. In 2004 it still had a dominant position in fixed and cable services. During the 1990s the firm made investments in eleven countries, including Angola, Mozambique or Guinea-Bissau where political and military instability was the rule. But these investments were made by following government instructions and the sums invested, and thus the consequent losses, were small and “hidden” by the huge revenue stream arising from the dominant position of the firm in the Portuguese market. This type of reasoning was usual among Portuguese firms where the government had a say in the strategy. From the sample of 112 operations, 20 may be partially influenced by a firm’s previous gains. Three other firms – CGD, EDP and Cimpor – benefited from a similar monopoly situation in the Portuguese market. CGD had the monopoly of banking for public servants in Portugal for more than twenty years. EDP was the sole provider of electricity. Cimpor was a monopolist in the cement industry. They were able to

2 As referred in Hosseini (2005), uncertainty explains why FDI flows occur more frequently among developed countries. 3 Normality tests confirm that it is not possible to assume a normal distribution in the population.

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

397

Table 3 Consolidated profits and ROA (values in Million Euros). Profits

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

CGD EDP PT

291.17 n.a. 180.83

248.45 n.a. 273.95

528.09 n.a. 349.64

350.75 522.79 441.1

349.13 513.94 494.68

544.47 548.97 540.32

653.78 450.83 307.39

665.13 335.22 391.05

667.25 381.11 240.23

448.48 440.15 500.12

ROA CGD EDP PT

0.85% n.a. 4.31%

0.68% n.a. 6.29%

1.28% n.a. 7.01%

0.72% 4.22% 4.75%

0.63% 3.75% 5.81%

0.87% 3.69% 4.09%

0.98% 2.78% 1.74%

1.00% 1.85% 2.85%

0.90% 2.04% 1.77%

0.64% 1.95% 3.86%

Source: Firms’ annual reports. ROA = profits/total assets.

absorb sufficient liquidity during the monopoly years that partially motivated and was later used to invest abroad. Table 3 shows the obtained profits since 1995 by three of these firms. It may be seen that the return on assets of EDP and PT decreases significantly after 2000 due to the liberalization of both industries. CGD, on the other hand, presented in the 1990s, on average, a higher financial margin (by 0.5%) than the banking sector due to access to cheaper funding. Interviews implicitly confirm a so-called house money effect (Thaler and Johnson, 1990). The manager of EDP stated that “the market was mature for us and the firm generated excessive cashflows for our needs in Portugal. Thus, we needed to invest abroad”. The same happened with CGD and PT. Moreover, the manager of Secil when explaining the internationalization policy of Cimpor: “they had a privileged situation during the privatization of the industry (in the 1990s) because the state left them with a lot of money to invest abroad”. This shows how consecutive Portuguese governments “allowed” state owned firms to earn money from their monopoly position and start the internationalization process before facing competition in the Portuguese market. In the above situations managers dealt with uncertainty by using a “mental accounting” rule where the willingness to accept a higher risk is explained by a lower, in comparison with prior gains, potential loss. This “mental accounting” was recurrently used and kept throughout the years, thus becoming a behavioural rule. Its economic consequence is that these decisions do not follow neoclassical predictions because managers, by keeping less or nonprofitable investments operating for a long time, are not acting as profit maximizers and investing up to the point where the marginal product of capital is just sufficient to cover the real user cost of capital. Table 4 shows that the main strategic investment abroad made by each firm was not only less profitable than the earnings in the home market but have accumulated losses for a long time. One might think of political intervention, and not of economic motifs, as the reason why managers made sub-optimal decisions and kept a non-profitable investment operating. Managers may have felt obliged to comply with governmental instructions in order to keep their jobs. However, mental accounting is not exclusive to government control. Salvador Caetano, a privately owned bus producer and car assembler, invested in the United Kingdom to channel domestic production. The investment in the UK started in 1984 when a local representative “convinced” the firm to invest in the country but the results were not satisfactory: “The level of profits was not good and we had several years of losses due to the negative

impact of tourism and the difficulties of tour operators. These invested in used buses and destroyed the market for new ones”. In 1998 the firm made a new investment in the UK to produce coaches with a local partner. But the agreement broke down because the partner decided to joint venture with another firm. Again, market reasons explained the failure: “. . .we lost a lot of money due to market context, namely the demand for coaches that changed after the new investment was made”. Finally, in 2004 the firm invested again in the production of buses in Portugal to export to the UK and closed the production of coaches by transforming it in a car repairing business. The manager justifies the continuing investment in the UK with the possibility of channelling Portuguese production. But the fact is that the firm has been investing continuously since 1984 without profiting from it and their managers were not able to learn from the different attempts to change the business. This investment has been financed mainly with domestic cash-flows from car assembly to Toyota. Thus, the ability to limit the losses and the expected possibility to recover it has influenced the investment. A different variable related with the past and playing a part in the location decision by Portuguese firms is anchoring due to cultural and historical links. One third (38 in 112) of the studied investments are located in Portuguese speaking countries or are explained by historical reasons. In these cases, the decision to invest abroad was evaluated from a particular starting point, common historical and cultural background, and the choice of this anchoring point influenced the location decision. The relevance of cultural variables is confirmed in the interviews. Seven firms explicitly stated that cultural variables were determinant for investment location and a further three also referred to their relevance. Moreover, six firms present historical reasons to be present in a market. The manager of Modelo, a retailer, explained the investment in Brazil in this way: “We were in Brazil since 1989 through a partnership (in an industrial area). So, when we decided to invest there we already knew the market. The cultural aspect and the special affinity of the CEO to the country were decisive in the choice of Brazil”. The manager of EDP stated that “Brazil was a natural market for us due to the opportunity (liberalization of the Brazilian market) and cultural reasons”. The available data shows that cultural and historically driven FDI is not necessarily an advantage for the firm. Table 5 presents information for 2004, when international operations were running for some years and thus had sufficient time to become profitable. There are, in the sample, sixteen locations with lower and eleven with higher return on assets than that of firms’ consolidated accounts.

Table 4 Net income of the main strategic investments abroad (values in Million Euros).

a

CGD EDP PT

Country

1997

1998

1999

2000

2001

2002

2003

2004

Spain Brazil Brazil

n.a. n.a. –

n.a. n.a 17.0

n.a. −24.9 9.4

n.a. 7.25 122.1

3.1 79.9 −519.0

1.8 −20.2 −34.2

0.4 −86.3 −9.88

−11.7 48.1 −59.2

Source: Firms’ annual reports. a CGD invested in Spain in 1991 and it kept on having losses for a decade (Pinheiro Alves, 2001).

398

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

Table 5 Relative performance of FDI in Portuguese speaking countries. Country Firm

Consolidated ROA (2004)

Angola

Brazil

Cape Verde

Guinea Bissau

Macao

Mozambique

S. Tomé Principe

Timor

BES BPI CGD EDP JM Mota PT SC Secil Modelo BCP

0.60% 0.80% 0.64% 1.95% 4.14% 1.70% 3.86% 1.40% 7.83% 5.15% 0.72%

Higher Higher – – – Higher n.a. Lower Lower – Lower

Lower – Lower Higher Lower – Lower – – Lower –

– – Lower Lower – – n.a. Higher Lower – –

– – – – – – Lower Lower – – –

Lower – Lower n.a. – – n.a. – – – Higher

– Higher Higher – – Higher n.a. Lower – – Higher

– – n.a. – – – n.a. – – – –

– – Higher – – – n.a. – – – –

Source: Firms’ annual reports.

The maintenance of non-profitable investments confirms that these decisions were not made in accordance with the maximization goal. Moreover, the effect of anchoring was to change the direction of FDI flows from locations with a potential higher return. 4.2. Present situation Situations in the present may also influence FDI location decisions. In 1996, the Portuguese government chose Brazil as the main target for the Portuguese economy (NPI, 1997). The Portuguese prime-minister at the time made several speeches and visits to this country, explicitly exhorting investors to move to that market. IPE (a state owned holding) participated as a shareholder in the investment made by a private firm. Furthermore, the year 2000 marked the 500th anniversary of the arrival of Pedro Àlvares Cabral to Brazil (following the 500th anniversary of Vasco da Gama’s journey to India), with celebrations both in Portugal and in Brazil. There was, on that period, a huge stream of news about the attractiveness and the opportunity of investing in Brazil. Portuguese firms were moving abroad and cultural ties, common language, a huge market and specific incentives such as interest free loans explained the sudden interest in Brazil (Costa, 2003). There were, in 2001, 147 investments in Brazil made by 83 parent Portuguese companies and a large majority of these had invested after 1996. Almost one in three Portuguese firms with investments abroad at the end of the 1990s chose the Brazilian market. The frequent news about firms investing in Brazil tended to reinforce that influence and to create a “herding” phenomenon through social learning. The manager of BES, a Portuguese bank, confirms it: “We went with other firms such as PT, JM and Sonae”. The same happened with a retail firm, Modelo: “We had a lot of cash to spend and the government had limited the number of licences to operate in Portugal. So, we decided to invest abroad. On the occasion Brazil and Latin America were the most fashionable locations and this (the investments) has a lot to do with fashions, as you know”. This behavioural similarity is understandable in a country like Portugal, where the managerial community is a small group where everybody knows each other and competitor’s moves or reputation concerns by those managers that do not “follow the herd” may influence decisions. The outcome of these investments was not always the expected. Only two of the operations included in our sample are still running and the remaining were sold out. The participation of the Portuguese state in a private enterprise was sold with a huge loss. Between 1997 and 2001 the Portuguese firms invested a total of 13,000 Million Euros in Brazil but divested half of this amount (Banco de Portugal, 2005). This indicates that a significant part of the investments were not successful and firms had to leave the market. Given the short period of time, and despite the losses, one cannot say that these were sub-optimal decisions because firms have tended to correct them. But the availability of news about

“re-discovering Brazil” led managers to overestimate this market and transfer their attention from other potential locations, thus influencing the size and the direction of FDI flows towards Brazil. Managers may also act in conformity with informal but socially accepted moral standards when making FDI decisions. Salvador Caetano, a bus producer and car assembler, has invested in Mozambique, in 1995, in a joint venture with the Mozambican government. The agreement did not include any written guarantee regarding the level of tariffs. According to the manager, the investment was influenced by an attitude of fairness towards a very poor country: “We wanted to help the development of Mozambique and agreed, with the state as a partner, to install a factory to produce components and assemble buses. But the government, instead of giving some type of protection to the industry, decided to raise tariffs for the import of components and to eliminate tariffs for the import of buses. Thus, the factory is now inactive because there are no necessary conditions to develop the business. And we are very disappointed. It seems that they do not want our help”. This manager reflects a common feeling in Portugal about the need to invest in the ex-colonies and help them to develop, and confirms that moral influences also have a specific role in economic decisions, namely in the direction of FDI flows. A further example is to invest in East Timor, a Portuguese excolony and a very poor country. The decolonization process was badly managed by Portugal resulting in an occupation by Indonesia between 1975 and 1999. Thus, there is a common will, in the Portuguese society, to help East Timor. PT, a telecommunications operator, invested in the newly independent state of East Timor in 2003 despite the absence of Asia in its international strategy. There are a total of 12 examples in the firms surveyed, namely those closely associated with the Portuguese policy of helping former colonies. In all situations the direction of FDI flows has changed from the prediction of the neoclassical model, for capital to look for the higher marginal rate of return, to reasons associated with fairness. 4.3. Future’s reference Banco Espirito Santo (BES), a bank, and Jerónimo Martins (JM), a retailer, decided to invest in the Brazilian market in the second half of the 1990s in order to increase revenue and profitability. However, optimistic expectations about the outcome of the investments led both managers to be overconfident and to disregard relevant information in their choice of the country. The group where BES is included is in the country since 1975, although in different business areas. The business experience in the country seemed to provide reassuring knowledge about the success of the new operation. This is implicitly confirmed by its manager: “We are in Brazil since 1975 (insurance, investment banking and agriculture businesses) and, in 1997, decided to buy a bank with a retail network. But it did not work that well because it is a

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

399

Table 6 Association tests: uncertainty and behavioural rules (“Numbehav”). Situation 1Uncertainty proxy: Symmetric Cramer’ Va Contingency coef.a Kendall’s tau-b Gamma Directional Lambda Goodman Kruskal Tau Somers’ d Eta square a ***

ValueCountryrating

N

Signific.

ValueTypeofcountry

N

Signific.

0.367 0.461 0.467 0.640

112 112 112 112

0.000*** 0.000*** 0.000*** 0.000***

0.496 0.574 0.273 0.344

112 112 112 112

0.000*** 0.000*** 0.000*** 0.000***

0.118 0.089 0.515 0.544

112 112 112 112

0.009*** 0.000*** 0.000*** –

0.221 0.169 0.280 0.685

112 112 112 112

0.009*** 0.000*** 0.000*** –

Requires aggregated data in “Numbehav” because they are based on the chi-square. Significant at a 1% level.

very peculiar market where foreigners are usually not successful. It is necessary to rely on local management because they know better the market”. However, despite the long business activity in Brazil and 20 years of accumulated knowledge of the market, the firm did not hire local managers to run its new business and relied on expatriates. Therefore, the investment of 1997 showed not only an inability to learn by the firm and its managers but also overconfidence on its own management ability to obtain different and better results than those of other banks (often with more international experience) that previously had failed in the market (Hilton, 2003). The manager recognized that up to 2004 only one foreign bank, ABN Amro, was able to be successful in the Brazilian market. Moreover, the manager relied on its previous successes in Brazil by accepting an illusory correlation that less developed markets are associated with higher and easier profitability. The investment of JM in Brazil was also influenced by overconfidence from its manager and main shareholder: “I was marketing manager of Unilever in Brazil. I knew the market . . . if I didn’t I would have not committed so many mistakes”. He reinforces by saying that “It was nonsense to go to Brazil. It is a very different market, with powerful competitors, both locals and foreigners, very strong and with a lot of money. We have no balance sheet for the market”. But the manager had information about competitors and knew that the need for a large balance sheet is a characteristic of the retail business. Therefore, an illusory perception about his abilities has influenced the decision. The manager recognizes: “due to a stupid pride I was convinced that we would make it”. In a similar way to other firms, these two MNE have decided the direction of FDI flows by following other criteria than simply to maximize the net present value of the difference between revenue and costs. 5. The role of uncertainty The Heiner model predicts a positive and significant relationship between uncertainty and the use of behavioural rules in FDI location decisions. In order to test it, a variable called “Numbehav” is formed from the number of rules of behaviour detected for each FDI operation (only 160, of a total of 175, are related to a specific location). For uncertainty, the most obvious way to measure it is sovereign risk ratings (proxy 1: “Countryrating”) as presented by historical ratings of Standard&Poor’s for long term debt in local currency (Annex A describes all the variables). This is not a perfect proxy because ratings strictly represent the ability of the country to pay its sovereign debt and do not include industry-specific market risk from the existing competition or the “unknown unknowns” of uncertainty. Nevertheless, the inclusion of different risks indicates that the “known unknowns” of uncertainty are, at least partially, considered. A second proxy is to define uncertainty by underlining the cultural connections of Portugal. Thus a second variable called “Typeofcountry” is also considered where Portuguese speaking countries are regarded as having more uncertainty than OECD

countries, to which Portugal is more integrated in economic, legal and political terms, and less uncertainty than the remaining countries with no special connections with Portugal. This emphasis on cultural connections is in agreement with the perception of managers. For example, the manager of Secil says, when justifying the presence in Angola and Cape Verde: “Our Irish partners do not understand the advantage of having a close cultural relationship with these countries”.4 Several independence and association tests were performed. Independence tests based on the chi-square statistic, which are needed given the weak linearity existing between the level of uncertainty and the number of behavioural rules, show that the null hypothesis can be rejected at a significance level of 1% meaning that lesser than 1 sample in 100 would show independent variables.5 After confirming non-independence, measures of association and direction were tested in order to understand how strongly the two variables are related. The results for “Countryrating” and “Typeofcountry” confirm the existence of a relationship between uncertainty and the number of behavioural rules as predicted by the Heiner model (Table 6). Most of the values are fairly strong and all measures indicate positive direction.6 The Eta coefficient, for example, shows that “countryrating” explains 54%, and “Typeofcountry” 68%, of the variability in “Numbehav”. These results cannot ascertain that behavioural rules are caused by higher uncertainty. Other reasons may explain the relationship between both variables. To check it, four control variables are considered: the influence of shareholders in decision-making; the managerial role; the stated goals of each firm; and the previous level of internationalization. For all the four control variables the association between uncertainty and behavioural rules is still significant (Table 2 in Annex B). Therefore, the association existing between the level of uncertainty and the number of behavioural rules detected in Portuguese firms indicates the Heiner model as a good tool to predict managers’ behaviour in FDI location decisions. The model predicts that managers, when facing uncertainty, rely on actions which are adaptable to relatively recurrent situations, such as investing where there is an historical or cultural tie, or when a fad signs to managers the place to invest, while disregarding actions which are appropriate in unusual circumstances. There were behavioural rules in 55% of the Portuguese FDI operations, mainly located in places ranked with higher uncertainty. The remaining operations, without behavioural

4 Two other proxies to uncertainty, arising from the experience of internationalization of each firm, were tested but without meaningful results. These proxies were based on the assumption that higher international experience would reduce the use of rules of behaviour, but a changing environment, and its effect on the perceptions of managers, may prevent learning experience to reduce their use. 5 Annex B, Table 1. 6 Measures based in proportional reduction in error present low values due to the use of more disaggregated data in comparison with the remaining measures.

400

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

Table 7 Type of behavioural rules in the Heiner model. Countryrating and Numbehav

Rules of extrinsic origin

Tests

Value

Deg. Freed

N

Significance

Value

Deg. Freed

N

Significance

23.5 26.4

4 4

112 112

0.000*** 0.000***

22.4 21.6

4 4

112 112

0.000*** 0.000***

0.324 0.417 0.396 0.568

– –

0.000*** 0.000*** 0.000*** 0.000***

0.317 0.409 0.353 0.560

– –

–

112 112 112 112

112 112 112 112

0.000*** 0.000*** 0.000*** 0.000***

0.150 0.094 0.418 0.466

– – – –

112 112 112 112

0.017** 0.000*** 0.000***

0.083 0.141 0.313 0.389

– – –

112 112 112 112

0.548 0.000*** 0.000***

Independence Pearson Chi-square Likelihood ration Symmetric Cramer’ V Contingency coef. Kendall’s tau-b Gamma Directional Lambda Goodman Kruskal Tau Somers’ d Eta square

Rules of intrinsic origin

–

Countryrating and Numbehav

Rules originated in the present

Tests

Value

Deg. Freed

N

Significance

Value

Deg. Freed

N

Significance

11.8 11.4

4 4

112 112

0.018** 0.022**

40.2 43.8

4 4

112 112

0.000*** 0.000***

– –

112 112 112 112

0.006*** 0.006*** 0.001*** 0.001***

0.446 0.533 0.503 0.711

– –

112 112 112 112

0.000*** 0.000*** 0.000*** 0.000***

112 112 112 112

a

0.200 0.206 0.491 0.586

– – –

112 112 112 112

0.017** 0.000*** 0.000***

Independence Pearson Chi-square Likelihood ration Symmetric Cramer’ V Contingency coef. Kendall’s tau-b Gamma Directional Lambda Goodman Kruskal Tau Somers’ d Eta square a * ** ***

0.284 0.373 0.276 0.421 – 0.041 0.282 0.304

– – – – –

Rules originated in the past

0.032** 0.001***

–

Cannot be computed because the asymptotic standard error equals zero. Significant at a 10% level. Significant at a 5% level. Significant at a 1% level.

rules, were located in more developed countries, where uncertainty is lower. It should be noted that, despite the predicted behavioural “rigidity”, the Heiner model or the existence of herding, anchoring and other biases does not imply that managers do not care about profitability. A manager is not going to invest abroad if he does not believe the investment to be profitable. But the presented evidence shows that perceptions may prevent the choice of more profitable alternative locations and thus fills the gap left by neoclassical economics in what concerns non-maximizing location decisions kept unchanged over the years. These heuristics are FDI determinants arising from the decision making process and not from the traditional neoclassical reasons behind the decision to locate production, such as market, asset or efficiency-seeking issues. Finally, a further issue is to assess the individual relevance of each type of behavioural rule, as defined in the taxonomy. There were identified 114 extrinsic and 46 intrinsic behavioural rules, and, in the time span, 73 related with the past and 84 with the present. Statistical tests show all types not to contradict the Heiner model, except rules of intrinsic nature in the second proxy where the results are very weak (Table 7). Rules originated in the past – anchoring, learning and mental accounting – are the ones presenting a stronger influence in Portuguese FDI location decisions. 6. Conclusions Behavioural rules do influence managers in information collection, the selection of alternatives and the final FDI location decision. The FDI determinants arising from the behavioural approach were helpful in influencing 55% of the locations chosen by Portuguese managers. There were several situations where managers repeated

the same behaviour in a consecutive way and kept unchanged nonmaximizing decisions for a long period. Among the identified types of regularities those originated in the past and of extrinsic nature seem to have a more significant role in influencing FDI location decisions by Portuguese firms. But those of intrinsic nature and originated in the present are also significant. The influence of behavioural rules in FDI decisions has two main economic consequences. First, by considering the environmental context of decision making and the perceptions and motivations of firms’ decision makers, it is possible to reach a better understanding of how FDI decisions deviate from the predicted neoclassical rationality. The evidence above presents examples of sub-optimal decisions, such as the operation of non-profitable investments for a long period, where firms do not equate marginal rates of return in their different investment locations. This is not the “satisficing” of the traditional behavioural theory of the firm, where constraints from the limited capabilities of the human being explain why MNE are not able to indefinitely improve their behaviour towards maximization. It is the following of behavioural rules such as fairness, herding or anchoring, that while apparently reducing uncertainty, lead to the overestimation of expected outcomes in FDI decisions and prevent the choice of alternative locations with a higher return. Second, the use of behavioural rules has often changed the predicted direction of FDI flows. Evidence shows that the perceptions of managers influence the direction of FDI flows to places where other managers are investing, to locations with great availability in the news or to countries where managers, firms or the society as a whole, had previous experiences. These “unpredicted” directions are not necessarily where the firm could expect to obtain the highest rate of return from their investments, as per the neoclassical model, but they are chosen nonetheless.

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

These results are based on Portuguese data and one can ask whether they are valid for other countries. The number of rules of behaviour for each of the characteristics (the time span where they originate or their intrinsic or extrinsic nature) is surely specific to the Portuguese case. But there are no obvious reasons why heuristics found in financial markets and in some FDI operations cannot be applied to FDI managers in different locations. Even the case of government intervention in FDI location decisions, where the direction of investment flows is oriented, is common to all countries. East Timor, for example, has received investment from Portugal, Australia and other countries. The most important is that the behavioural approach brings in new variables to FDI location decisions that are not solely based on the will to increase revenue or reduce costs and are extensible to investors of different nationalities when applying simplifying strategies as a way to deal with uncertainty. Finally, the findings and consequences presented above rely on some features that may be improved in future work. First, behavioural rules are complimentary in the explanation of FDI both to the neoclassical approach, by relaxing the assumption that investors are rational decision makers that behave as if maximizing by choosing the best alternative from a set of options based on probabilistic risk adjusted expected returns, and to the traditional behavioural theory of the firm. Given the examples of behavioural “rigidity”, future research should improve the knowledge about the way FDI decisions are made. Moreover, although heuristics indicate that agents are not able to learn from past experiences in all situations and thus cannot improve indefinitely their behaviour towards optimality, the knowledge about behaviour rules might help managers improving their performance by considering and reviewing their use. Second, the behavioural approach may be deepened through a better understanding of the role of contextual issues such as the influence of each firm’s culture and history and the individual cognitive characteristics of managers, including cultural and moral variables. This suggests the existence of further behavioural influences in FDI operations. Acknowledgements This paper is part of PhD research financed by Fundac¸ão para a Ciência e a Tecnologia and Quadro Comunitário de Apoio III. I am grateful to Prof. Philip Jones and John Cullis, at the University of Bath, and to three referees for comments. The usual disclaimer applies.

401

Table B.1 Independence tests. Independence tests

Countryrating and Numbehav Value

DFr

N

Significance

30.2 34.9

4 4

112 112

0.000*** 0.000***

Typeofcountry and Numbehav Pearson chi-square 55.0 Likelihood Ratio 64.4

4 4

112 112

0.000*** 0.000***

Pearson chi-square Likelihood Ratio

***

Significant at a 1% level.

different groupings so that their robustness can be checked. The first considers three sets of country locations with zero rules, 1 rule or 2 rules or more. The second considers zero rules, 1 or 2 rules, and 3 rules or more. Uncertainty: Countryrating: Rating of the country where FDI is located. A means lower, B intermediate and C higher uncertainty (risk). Since these are not available for all cases, and given that the level of development is usually (by rating firms) recognized to be negatively correlated with risk, they are replaced, when absent, by the measure of development used by the World Bank.Typeofcountry: Divided by: Countries with a similar law and political and economic institutions (OECD and EU) where there is less uncertainty; countries with a common tongue and past with Portugal; remaining countries, with more uncertainty. Control variables: Decision: Influence of shareholders in decision-making. The shareholder structure did not significantly change in the past for the considered firms. This is divided in 4 categories: Individual decisions with more than 5% and less than 50% or more than 50%. And group decisions when the firm is public or when the Portuguese government has a role. Respondents: Influence of respondents divided in 3 categories: CEO’s, Other members of the board and Middle managers. Objective: Stated goals of the firm divided in 5 categories: Maximization, Minimum profitability, Other quantitative objective, Qualitative objectives and at least two of the last three. Previlevel: Previous level of internationalization is classified in 3 different categories: (a) Lower level (of internationalization): when the firm only has investments abroad less than 5 years old; (b) Medium level: when the firm has FDI for 5 or more years but it is present in less than 5 countries: (c) Higher level: when the firm has FDI for 5 or more years and it is present in more than 5 countries.

Annex A. Description of variables in statistical tests Annex B. Rules of behaviour: Numbehav: Number of behavioural rules. When aggregated information is required tests are made for two

See Table B.1 and Tables B.2a–B.2d

Table B.2a Countryrating” and behavioural rules—tests for control variables (Decision). Control variable: Decision

1: Individual more than 50%

2: Individual more than 5%

3: Group in public firm

4: Group with State role

Situtation 1

Value

N

Significance

Value

N

Significance

Value

N

Significance

Value

N

Significance

0.320 0.459

37 37

0.013** 0.013**

0.583 0.857

13 13

0.004*** 0.004***

0.432 0.818

25 25

0.037** 0.037**

0.541 0.682

37 37

0.000*** 0.000***

0.174 0.071 0.329 0.342

37 37 37 37

0.278 0.246 0.013**

0.286 0.289 0.750 0.911

13 13 13 13

0.124 0.129 0.004***

0.400 0.300 0.519 0.548

25 25 25 25

0.029** 0.027** 0.037**

0.179 0.127 0.608 0.620

37 37 37 37

0.016** 0.004*** 0.000***

N Total = 112 Symmetric Kendall’s tau-b Gamma Asymmetric Lambda Goodman and Kruskal Tau Somers’ d Eta * Significant at a 10% level. Significant at a 5% level. *** Significant at a 1% level. **

402

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403

Table B.2b Countryrating” and behavioural rules—tests for control variables (Respondents). Control variable: Respondent

N Total = 112 Symmetric Kendall’s tau-b Gamma Asymmetric Lambda Goodman and Kruskal Tau Somers’ d Eta a * ** ***

1: CEO

2: Member of the Board

3: Middle Manager

Value

N

Significance

Value

N

Significance

Value

N

Significance

0.617 1.000

5 5

0.025** 0.025**

0.517 0.742

47 47

0.000*** 0.000***

0.465 0.619

60 60

0.000*** 0.000***

0 0.167 0.667 0.662

– 5 5 5

a

0.513 0.025**

0.321 0.165 0.568 0.569

47 47 47 47

0.003*** 0.000*** 0.000***

0.111 0.103 0.510 0.579

60 60 60 60

0.095* 0.000*** 0.000***

Cannot be computed because the asymptotic standard error equals zero. Significant at a 10% level. Significant at a 5% level. Significant at a 1% level.

Table B.2c Countryrating” and behavioural rules—tests for control variables (Objective). Control variable: Objective

N Total = 112 Symmetric Kendall’s tau-b Gamma Asymmetric Lambda Goodman and Kruskal Tau Somers’ d Eta * ** ***

1: Minimum profitability

2: Other quantitative objective

3: Qualitative objectives

4: 1, 2 and 3 together

Value

N

Significance

Value

N

Significance

Value

Value

N

Significance

0.483 0.687

20 20

0.001*** 0.001***

0.567 0.752

47 47

0.000*** 0.000***

0.399 0.568

44 44

0.001*** 0.001***

0.273 0.225 0.528 0.528

20 20 20 20

0.060* 0.074* 0.001***

0.161 0.123 0.610 0.625

47 47 47 47

0.122 0.002*** 0.000***

0.040 0.066 0.444 0.523

44 44 44 44

0.562 0.137 0.001***

N

Significance

Only 1 case

Significant at a 10% level. Significant at a 5% level. Significant at a 1% level.

Table B.2d Countryrating” and behavioural rules—tests for control variables (Previlevel). Control variable: Level Internationalization

N Total = 107 Symmetric Kendall’s tau-b Gamma Asymmetric Lambda Goodman and Kruskal Tau Somers’ d Eta

1: Lower level

2: Medium level

3: Higher level

Value

N

Significance

Value

N

Significance

Value

N

Significance

0.308 0.446

41 41

0.003*** 0.003***

0.529 0.698

27 27

0.000*** 0.000***

0.563 0.747

39 39

0.000*** 0.000***

0.192 0.101 0.353 0.370

41 41 41 41

0.017** 0.027** 0.003***

0.167 0.143 0.573 0.544

27 27 27 27

0.245 0.034** 0.000***

0.087 0.142 0.616 0.648

39 39 39 39

0.311 0.001*** 0.000***

* Significant at a 1% level. Significant at a 5% level. *** Significant at a 10% level. **

References Aharoni, Y., 1999. The foreign investment decision process. In: Buckley, P.J., Ghami, P.N. (Eds.), The Internationalization of the Firm. A Reader. ITPB, London. Arrow, K., 1996. In: Kenneth Arrow, Enrico Colombatto, Mark Palman, Christian Schmitt (Eds.), The General Notion of Rationality, Preface to “The Rational Foundations of Economic Behaviour”. St. Martin’s Press, London. Banco de Portugal, 2005. International Investment Statistics., http://www.bportugal.pt/. Banerjee, A.V., 1992. A Simple Model of Herd Behavior. The Quarterly Journal of Economics 107 (August (3)), 797–817. Beckmann, D., Lukas, M., Megumi, S., 2008. Does culture affect asset managers’ views and behavior? Journal of Economic Behavior & Organization, doi:10.1016/j.jebo.2007.12.001. Blonigen, B., 2005. A Review of the Empirical Literature on FDI Determinants, NBER WP nr. 11299. Brainard, L., 1993. An Empirical Assessment of the Factor Proportions Explanation of Multinational Sales, Working Paper nr. 3624-EFA, Sloan School of Management, MIT.

Caves, R., 1996. Multinational Enterprise and Economic Analysis, 2nd ed. Cambridge University Press, Cambridge. Costa, C.G., 2003. Portuguese Investment in Brazil: Traditional Determinants and the Contribution of Cultural Proximity. PhD Dissertation. Instituto Superior de Economia e Gestão, Lisbon. Frey, B., Eichenberger, R., 2001. Economic incentives transform psychological anomalies. In: Inspiring Economics. Bruno Frey, Edward Elgar, Cheltenham, pp. 21–36. Grinblatt, M., Keloharju, M., 2001. How distance, language, and culture influence stockholdings and trades. The Journal of Finance LVI (June (3)), 1053–1073. Heijdra, B., 1988. Neoclassical economics and the psychology of risk and uncertainty. In: Peter Earl (Ed.), Psychological Economics. Kluwer Academic Publishers, Lancaster. Heiner Ronald, 1983. The origin of predictable behavior. American Economic Review 73 (September (4)), 560–595. Heiner, R., 1985. Origin of predictable behavior: further modeling and applications. In: American Economic Review, AEA papers and Proceedings, May, 75, pp. 391–396. Heiner, R., 1989. The origin of predictable dynamic behavior. Journal of Economic Behavior and Organization 12, 233–257.

R. Pinheiro-Alves / The Journal of Socio-Economics 40 (2011) 394–403 Hilton, D.J., 2003. Psychology and the financial markets: applications to understanding and remedying irrational decision-making. In: Brocas, I., Carrillo, J.D. (Eds.), The Psychology of Economic Decision, vol. II. Oxford University Press, pp. 273–297. Horstmann, I., Markusen, J., 1992. Endogenous market structures in international trade (Natura Facit Saltum). Journal of International Economics 32 (February), 109–129. Hosseini, H., 2005. An economic theory of FDI: a behavioral economics and historical approach. The Journal of Socio-Economics 34, 528–541. Johanson, J., Vahlne, J., 1977. The internationalization process of the firm—a model of knowledge development and increasing market commitments. Journal of International Business Studies 8 (1), 23–32. Kahneman, D., Knetch, J.L., Thaler, R.H., 1986. Fairness as a constraint on profit seeking: entitlements in the market. American Economic Review 76 (4), 728–741. Katona, G., 1975. Psychological Economics. Elsevier, New York. Kinoshita, Y., Mody, A., 2001. Private information for foreign investment in emerging economies. Canadian Journal of Economics 34 (May (2)), 448–464. Knight, F., 1921. Risk, Uncertainty and Profit. Houghton Mifflin, Boston. Malmendier, U., Tate, G., 2005. CEO overconfidence and corporate investment. The Journal of Finance LX (December (6)), 2661–2700. Neuendorf, K., 2002. The Content Analysis Guidebook. Sage Publications, London.

403

Norusis, M., 2003. SPSS 12.0 Statistical Procedures Companion. Prentice Hall, New Jersey. NPI, 1997. New Internationalization Policy, Decision of the Ministerial Council nr 61/97, 15 April, Presidency of the Ministerial Council, Lisbon. Pinheiro Alves, R., 2001. Portugal: Iberian Relations in view of Globalization and Market Liberalization. Ordem dos Economistas and Portuguese Ministry of Economy, Lisbon. Rabin, M., 1998. Psychology and economics. Journal of Economic Literature 36 (March (1)), 11–46. Shiller, R.J., 2003. From efficient markets theory to behavioral finance. Journal of Economic Perspectives 17 (Winter (1)), 83–104. Thaler, R.H., Johnson, E.J., 1990. Gambling with the house money and trying to break even: the effects of prior outcomes on risky choice. Management Science 36 (June (6)), 643–660. Tversky, A., Kahenman, D., 1982. Judgements of and by representativeness. In: Kahneman, D., Slovic, P., Tversky, A. (Eds.), Judgement under Uncertainty: Heuristics and Biases. Cambridge University Press, pp. 84–98. Williamson, O., 1981. The modern corporation: origins, evolution, attributes. Journal of Economic Literature 19, 1537–1568. Zwiebel, J., 1995. Corporate conservatism and relative compensation. Journal of Political Economy 103 (1), 1–25.