Chapter 8 Location Choice of Japanese Multinationals: Country Characteristics or Region Characteristics?

In the present global era in which firms choose the location of their plants beyond national borders, location characteristics are important for attracting multinational enterprises (MNEs). The better access to countries with large market is clearly attractive for MNEs. For example, special treatments on tariffs such as the Generalized System of Preferences (GSP) are beneficial for MNEs whose home country does not have such treatments. Not only such country characteristics but also region characteristics (i.e. province-level or city-level ones) matter, particularly in the case that location characteristics differ widely between a nation's regions. The existence of industrial concentration, that is, agglomeration, is a typical regional characteristic. It is with consideration of these country-level and region-level characteristics that MNEs decide their location abroad. A large number of academic studies have investigated in what kinds of countries MNEs locate, i.e. location choice analysis. Employing the usual new economic geography model (i.e. constant elasticity of substitution (CES) utility function, Dixit-Stiglitz monopolistic competition, and ice-berg trade costs), the literature derives the profit function, of which coefficients are estimated using maximum likelihood procedures. Recent studies are as follows: Head, Rise, and Swenson (1999) for Japanese MNEs in the US; Belderbos and Carree (2002) for Japanese MNEs in China; Head and Mayer (2004) for Japanese MNEs in Europe; Disdier and Mayer (2004) for French MNEs in Europe; Castellani and Zanfei (2004) for large MNEs worldwide; Mayer, Mejean, and Nefussi (2007) for French MNEs worldwide; Crozet, Mayer, and Mucchielli (2004) for MNEs in France; and Basile, Castellani, and Zanfei (2008) for MNEs in Europe. At the present time, three main topics can be found in this literature. The first introduces various location elements as independent variables. The above-mentioned new economic geography model usually yields the profit function, which is a function of market size, productive factor prices, price of intermediate goods, and trade costs. As a proxy for the price of intermediate goods, the measure of agglomeration is often used, particularly the number of manufacturing firms. Some studies employ more disaggregated numbers of manufacturing firms, such as the number of manufacturing firms with the same nationality as the firms choosing the location (e.g., Head et al., 1999; Crozet et al., 2004) or the number of firms belonging to the same firm group (e.g., Belderbos and Carree, 2002). As part of trade costs, some investment climate measures have been examined: free trade zones in the US (Head et al., 1999), special economic zones and opening coastal cities in China (Belderbos and Carree, 2002), and Objective 1 structural funds and cohesion funds in Europe (Basile et al., 2008). Second, the validity of proxy variables for location elements is further examined. Head and Mayer (2004) examine the validity of market potential on location choice. They propose the use of two measures: the Harris market potential index (Harris, 1954) and the Krugman-type index used in Redding and Venables (2004). The Harris-type index is simply the sum of distance-weighted real GDP. They employ the Krugman-type market potential index, which is directly derived from the new economic geography model, as it takes into account the extent of competition (i.e. price index) and is constructed using estimators of importing country dummy variables in the well-known gravity equation, as in Redding and Venables (2004). They find that "theory does not pay", in the sense that the Harris market potential outperforms Krugman's market potential in both the magnitude of its coefficient and the fit of the model to be estimated. The third topic explores the substitution of location by examining inclusive values in the nested-logit model. For example, using firm-level data on French investments both in France and abroad over the 1992-2002 period, Mayer et al. (2007) investigate the determinants of location choice and assess empirically whether the domestic economy has been losing attractiveness over the recent period or not. The estimated coefficient for inclusive value is strongly significant and near unity, indicating that the national economy is not different from the rest of the world in terms of substitution patterns. Similarly, Disdier and Mayer (2004) investigate whether French MNEs consider Western and Eastern Europe as two distinct groups of potential host countries by examining the coefficient for the inclusive value in nested-logit estimation. They confirm the relevance of an East-West structure in the country location decision and furthermore show that this relevance decreases over time. The purpose of this paper is to investigate the location choice of Japanese MNEs in Thailand, Cambodia, Laos, Myanmar, and Vietnam, and is closely related to the third topic mentioned above. By examining region-level location choice with the nested-logit model, I investigate the relative importance of not only country characteristics but also region characteristics. Such investigation is invaluable particularly in the case of location choice in those five countries: industrialization remains immature in those countries which have not yet succeeded in attracting enough MNEs, and as a result, it is expected that there are not yet crucial regional variations for MNEs within such a nation, meaning the country characteristics are still relatively important to attract MNEs. To illustrate, in the case of Cambodia and Laos, one of the crucial elements for Japanese MNEs would be that LDC preferential tariff schemes are available for exports from Cambodia and Laos. On the other hand, in the case of Thailand and Vietnam, which have accepted a relatively large number of MNEs and thus raised the extent of regional inequality, regional characteristics such as the existence of agglomeration would become important elements in location choice. Our sample countries seem, therefore, to offer rich variations for analyzing the relative importance between country characteristics and region characteristics. Our empirical strategy has a further advantage. As in the third topic in the location choice literature, the use of the nested-logit model enables us to examine substitution patterns between country-based and region-based location decisions by MNEs in the concerned countries. For example, it is possible to investigate empirically whether Japanese multinational firms consider Thailand/Vietnam and the other three countries as two distinct groups of potential host countries, by examining the inclusive value parameters in nested-logit estimation. In particular, our sample countries all experienced dramatic changes in, for example, economic growth or trade costs reduction during the sample period. Thus, we will find the dramatic dynamics of such substitution patterns. Our rigorous analysis of the relative importance between country characteristics and region characteristics is invaluable from the viewpoint of policy implications. First, while the former characteristics should be improved mainly by central government in each country, there is sometimes room for the improvement of the latter characteristics by even local governments or smaller institutions such as private agencies. Consequently, it becomes important for these smaller institutions to know just how crucial the improvement of region characteristics is for attracting foreign companies. Second, as economies grow, country characteristics become similar among countries. For example, the LCD preferential tariff schemes are available only when a country is less developed. Therefore, it is important particularly for the least developed countries to know what kinds of regional characteristics become important following economic growth; in other words, after their country characteristics become similar to those of the more developed countries. I also incorporate one important characteristic of MNEs, namely, productivity. The well-known Helpman-Melitz-Yeaple model indicates that only firms with higher productivity can afford overseas entry (Helpman et al., 2004). Beyond this argument, there may be some differences in MNEs' productivity among our sample countries and regions. Such differences are important from the viewpoint of "spillover effects" from MNEs, which are one of the most important results for host countries in accepting their entry. The spillover effects are that the presence of inward foreign direct investment (FDI) aises domestic firms' productivity through various channels such as imitation. Such positive effects might be larger in areas with more productive MNEs. Therefore, it becomes important for host countries to know how much productive firms are likely to invest in them. The rest of this paper is organized as follows. Section 2 takes a brief look at the worldwide distribution of Japanese overseas affiliates. Section 3 provides an empirical model to examine their location choice, and lastly, we discuss future works to estimate our model.

Regularization for sparsity in statistical analysis and machine learning

Pragmatism is the leading motivation of regularization. We can understand regularization as a modification of the maximum-likelihood estimator so that a reasonable answer could be given in an unstable or ill-posed situation. To mention some typical examples, this happens when fitting parametric or non-parametric models with more parameters than data or when estimating large covariance matrices. Regularization is usually used, in addition, to improve the bias-variance tradeoff of an estimation. Then, the definition of regularization is quite general, and, although the introduction of a penalty is probably the most popular type, it is just one out of multiple forms of regularization. In this dissertation, we focus on the applications of regularization for obtaining sparse or parsimonious representations, where only a subset of the inputs is used. A particular form of regularization, L1-regularization, plays a key role for reaching sparsity. Most of the contributions presented here revolve around L1-regularization, although other forms of regularization are explored (also pursuing sparsity in some sense). In addition to present a compact review of L1-regularization and its applications in statistical and machine learning, we devise methodology for regression, supervised classification and structure induction of graphical models. Within the regression paradigm, we focus on kernel smoothing learning, proposing techniques for kernel design that are suitable for high dimensional settings and sparse regression functions. We also present an application of regularized regression techniques for modeling the response of biological neurons. Supervised classification advances deal, on the one hand, with the application of regularization for obtaining a na¨ıve Bayes classifier and, on the other hand, with a novel algorithm for brain-computer interface design that uses group regularization in an efficient manner. Finally, we present a heuristic for inducing structures of Gaussian Bayesian networks using L1-regularization as a filter. El pragmatismo es la principal motivación de la regularización. Podemos entender la regularización como una modificación del estimador de máxima verosimilitud, de tal manera que se pueda dar una respuesta cuando la configuración del problema es inestable. A modo de ejemplo, podemos mencionar el ajuste de modelos paramétricos o no paramétricos cuando hay más parámetros que casos en el conjunto de datos, o la estimación de grandes matrices de covarianzas. Se suele recurrir a la regularización, además, para mejorar el compromiso sesgo-varianza en una estimación. Por tanto, la definición de regularización es muy general y, aunque la introducción de una función de penalización es probablemente el método más popular, éste es sólo uno de entre varias posibilidades. En esta tesis se ha trabajado en aplicaciones de regularización para obtener representaciones dispersas, donde sólo se usa un subconjunto de las entradas. En particular, la regularización L1 juega un papel clave en la búsqueda de dicha dispersión. La mayor parte de las contribuciones presentadas en la tesis giran alrededor de la regularización L1, aunque también se exploran otras formas de regularización (que igualmente persiguen un modelo disperso). Además de presentar una revisión de la regularización L1 y sus aplicaciones en estadística y aprendizaje de máquina, se ha desarrollado metodología para regresión, clasificación supervisada y aprendizaje de estructura en modelos gráficos. Dentro de la regresión, se ha trabajado principalmente en métodos de regresión local, proponiendo técnicas de diseño del kernel que sean adecuadas a configuraciones de alta dimensionalidad y funciones de regresión dispersas. También se presenta una aplicación de las técnicas de regresión regularizada para modelar la respuesta de neuronas reales. Los avances en clasificación supervisada tratan, por una parte, con el uso de regularización para obtener un clasificador naive Bayes y, por otra parte, con el desarrollo de un algoritmo que usa regularización por grupos de una manera eficiente y que se ha aplicado al diseño de interfaces cerebromáquina. Finalmente, se presenta una heurística para inducir la estructura de redes Bayesianas Gaussianas usando regularización L1 a modo de filtro.

An approach to operational modal analysis using the expectation maximization algorithm

This paper presents the Expectation Maximization algorithm (EM) applied to operational modal analysis of structures. The EM algorithm is a general-purpose method for maximum likelihood estimation (MLE) that in this work is used to estimate state space models. As it is well known, the MLE enjoys some optimal properties from a statistical point of view, which make it very attractive in practice. However, the EM algorithm has two main drawbacks: its slow convergence and the dependence of the solution on the initial values used. This paper proposes two different strategies to choose initial values for the EM algorithm when used for operational modal analysis: to begin with the parameters estimated by Stochastic Subspace Identification method (SSI) and to start using random points. The effectiveness of the proposed identification method has been evaluated through numerical simulation and measured vibration data in the context of a benchmark problem. Modal parameters (natural frequencies, damping ratios and mode shapes) of the benchmark structure have been estimated using SSI and the EM algorithm. On the whole, the results show that the application of the EM algorithm starting from the solution given by SSI is very useful to identify the vibration modes of a structure, discarding the spurious modes that appear in high order models and discovering other hidden modes. Similar results are obtained using random starting values, although this strategy allows us to analyze the solution of several starting points what overcome the dependence on the initial values used.