Metallurgy and materials Metalurgia e materiais Interconnecting financial performance and internationalization: a case of a rare metal producer

The paper aims to outline the causes that affected a rare earth metal producer Molycorp Silmet’s internationalization based on case study evidence. Moreover, based on statistical analysis, it aims to study how its internationalization and financial performance were interconnected. It concludes that 1) its export fluctuations mostly re-sulted from external factors and that despite such fluctuations, the firm’s financial health was mostly good, 2) high concentration on a single target market can lower profitability, 3) fluctuations in target market shares can increase or decrease a firm’s profitability, but they are not interconnected with its bankruptcy probability.


Introduction
Rare earth metal (REM) reserves have been discovered in 34 countriesfrom these, 15 in Asia and Australasia, 10 in Africa, six in Europe and three in America (Chen 2011). Most rare earth metals are produced in China, but it started protecting its environment via export quotas and taxes, and since summer 2015, resource taxes instead of them, which affected prices (Mancheri 2015;Massari & Ruberti 2013). Thus, many producers from other countries decided to increase exports (Biedermann 2014;Chen 2011) as REMs are critical for hi-tech and several other producers due to lack of substitutes (Golev et al. 2014;Klinger 2015). Considerable fluctuations in supply and demand can have a remarkable impact on firms' financial performance, especially for those operating outside China and mostly focusing on export markets. Therefore, it is very important to study the interconnections between the internationalization of firms -or in a nar-rower sense, target market choices -and their financial performance.
The best option for such research would be to study those firms that sell most of their products internationally and as such face high price fluctuations. Thus, this study focuses on such a REM manufacturer. Derived from the above, the study has two objectives. First, based on case study evidence, it aims to outline the causes that affected a REM producer's internationalization. Second, based on statistical analysis, it aims to study how its internationalization and financial performance were interconnected. The study focuses on an Estonian REM producer Molycorp Silmet that produces around 1/5 of the world's niobium and 1/10 of tantalum and exports almost all of its production.
Following the two objectives, the literature review section outlines two streams of literature: internationalization literature and literature on the financial performance of firms and its measurement. The empirical analysis presented herein starts from causes of internationalization based on case study evidence and, thereafter, a statistical analysis is conducted to study the interconnection between the case firm's financial performance and internationalization activities. The paper ends with conclusions, together with managerial and research implications.
The study makes several contributions. First, the causes of nonlinear internationalization -considerable fluctuations in exports by countries -have been rarely studied (Vissak 2014;Vissak & Francioni 2013). Second, it has not been studied if changes in shares of major export market groups affect the financial performance of REM producers. Finally, the study helps to understand what challenges a high-technology firm producing almost unique products must face due to global market forces.

Literature review
Although internationalization processes have been actively studied since the 1970s, we can still agree with Mtigwe (2006: 5) that "to date there appears to be no universally accepted model of international business, let alone the same theory of international business" and with Welch & Paavilainen-Mäntymäki (2014: 2) that a majority of studies on internationalization processes "have not taken a processual approach that incorporates time, dynamism and longitudinal observations." International business scholars have mostly focused on increasing involve-ment in internationalization and studied two types of internationalizers: 1) slow/ incremental which start from exporting to a few close countries and continue with farther markets and more complicated operation modes and 2) fast, 'born globals' that enter distant markets since their foundation. Still, it is not enough to focus only on growth. For instance, Nadkarni et al. (2006: 139) stated: "the international environment presents one of the toughest managerial challenges /…/ This is evident in the widespread internationalization failures around the world in terms of slow speed of international entries, withdrawals from foreign markets, divestments, and closure of foreign operations". As a result, recently, more attention has been paid to 3) nonlinear internationalization -"market exit and/or export decrease from any market by 25% or more" (Vissak & Masso 2015: 661). Thus, in Table  1, a short overview of three streams of internationalization research -slow, fast and nonlinear -is given and some factors affecting such internationalization paths are brought out.

Stream of literature A typical internationalization process
Factors leading to such an internationalization process the Uppsala (U-) model (Johanson & Vahlne 1977, 1990, Johanson & Wiedersheim-Paul 1975, innovation-related internationalization (I-) models (Bilkey 1978 fluctuating: firms experience fluctuations in international involvement: they exit some markets permanently or temporarily or decrease their involvement there (de-internationalize), later some re-enter some markets (re-internationalize) or enter others; such fluctuations can occur several times external and internal factors: including changed external environment (exchange rates, costs, competition, demand, export quotas), changes in business partners' situation, unstable foreign orders, lost or gained access to resources, managers' changes in the firm's strategy Table 1 An overview of research on internationalization processes.
REM producers' internationalization paths have not yet attracted attention in the literature, so it is not possible to find studies on how many firms belong to each of the above categories but some factors affecting their exports have been brought out. For instance, Biedermann (2014), Golev et al. (2014), Mancheri (2015) and Massari & Ruberti (2013) stated that due to export controls imposed by China, Chinese producers could not export as much as they wanted to, while producers in some other countries decided to export more as prices increased.
There is a myriad of literature on the financial performance of firms and it has been measured in a multitude of ways. The classical approach -a univariate analysis -uses single indicators (see Balcaen & Ooghe 2006;Dimitras et al. 1996). For that purpose, financial ratios indicating the profitability of assets (ROA) or sales (ROS) have been especially popular. Although ROA has been widely used, it has several limitations (for instance, it is not fully clear which assets and periods to include in the ratio denominator) and thus, ROS should be preferred when analyzing performance changes. In the calculation of ROS, a cash flow based ROS should be preferred over an accrual based formula, as cash flows can provide a more realistic picture of a firm's performance and its possible forthcoming failures (Laitinen & Laitinen 1998).
A more complex way to measure firms' performance is by using bankruptcy models. A bankruptcy model results in a score, bankruptcy probability or a binary outcome of survival or failure (Altman & Narayanan 1997;Dimitras et al. 1996;Bellovary et al. 2007). Of these, models resulting in a bankruptcy probability can be considered the best option, as they allow the comparison of a firm's performance dynamically on a unified scale (for instance, from 0% to 100% or from 0 to 1). Classically, binary logistic regression has been used for the creation of probabilistic bankruptcy models.
Although the production sector has been the most prominent research object for bankruptcy prediction scholars, no well-known models exist that would have been specifically composed for firms manufacturing metals (although some of such firms have been included in samples among other industrial firms). Thus, literature focusing on models created for manufacturing firms should be applied. The most prominent logit models are those provided in the pioneering study by Ohlson (1980), whose study encompassed three models to measure bankruptcy probability at three different time points. The models created by Ohlson have very high prediction abilities (ranging between 93 and 96%) and have rarely been outperformed by other logit models (see Appendix A in Bellovary et al. 2007). Thus, we will use this classical model in this study.

Method
This article is based on case study evidence as this method is suitable for studying causes of nonlinear internationalization processes (Vissak 2014; Vissak & Francioni 2013) and changes in financial performance. It allows researchers to understand how and why something happened, combine new empirical insights with previously developed theories and investigate complex phenomena taking into account their real-life context (Eisenhardt 1989;Yin 1994;Welch et al. 2011). Our data analysis followed three phases suggested by Miles & Huberman (1994): 1) organizing the data through discarding irrelevant data and writing summaries, 2) creating tables and figures for making conclusions and 3) collecting further data for verifying the initial conclusions. We used archival records, annual reports (from 2000-2014) and other materials as data sources.
We measured Molycorp Silmet's performance with two different indicators. First, we applied univariate analysis with a profitability ratio. We used one of the most common profitability measures to measure financial performance: operating cash flow divided by net sales (CF/S). As for most of the years, the firm did not have nonperforming accounts receivables and its amount of accounts receivables has not fluctuated considerably, we did not convert the sales measure in the denominator of cash flow based profitability ratio; instead we applied an accrual based measure (net sales from the income statement). From the three bankruptcy models created by Ohlson (1980), we applied the first model which measures the probability of bankruptcy one year before the event. It should be noted that in the case of Ohlson's bankruptcy model, values are in the range of 0 and 1, where 1 denotes the 100% likelihood of bankruptcy.
The case firm's nonlinear internationalization (specifically, its sales dynamics at different target markets) was reflected by the following measures. Firstly, the target market proportions were brought out for three market groups: the European Union as the firm's home region, the United States as its current owner's home market and the rest of the world (mostly South-East Asia). Such a distribution of markets also reflects the aggregate breakdown through different continents. Also, the firm's aggregate turnover over all target markets was used in the statistical analysis. Three variables were calculated to measure sales dispersion over different international target market groups. These variables are: a) standard deviation of the proportions of the above-mentioned target market groups, b) difference between the proportion of the largest and the smallest target market group, c) the proportion held by the target market group having the largest share.
Both nonlinear internationalization and performance measures were calculated for the period of 1999-2014, as earlier data were not publicly available. The interconnection between internationalization and financial performance measures was studied with Spearman's rank correlation coefficients. As the changes in Molycorp Silmet's internationalization behavior can alter its performance with a time lag, we found correlations with export variables from period t and performance indicators from periods t and t+1.

Results and discussion
The firm's predecessor was founded in 1927 but then it processed oil shale. It was nationalized in 1941 and destroyed during the World War II similarly to several other enterprises located in Estonia (Vissak 2014). In 1946 it started producing uranium oxide (production was ceased in 1990) and in 1970, rare earth metals and rare metals. In Soviet time it had 5500 employees. In 1997, the firm was fully restructured and privatized; then it had 1600 employees. Thereafter, ownership changed several times. The current owner Molycorp (operating in 10 countries and 25 locations) acquired it in April 2011. Silmet wished to increase its capacity for producing rare earth oxides (REO) and rare earth metals -niobium and tantalum (its main products) but also cerium, lanthanum, didymium, praseodymium and others. It also needed Molycorp's raw materials while the latter needed access to the European market. Molycorp Silmet annually produces up to 700 t of rare metals and 3,000 t of REO.  Molycorp Silmet's nonlinear internationalization -fluctuating foreign sales (see Figure 1) -was caused by several factors (see Table 2): changes in supply and demand, resource costs, prices and exchange rates, but it was also affected by owner-ship changes, Molycorp Silmet's efforts to increase its competitiveness and by Chinese economic policies (Golev et al. 2014;Massari & Ruberti 2013). Most of these factors were external, while in internationalization literature, also several internal factors -for instance, knowledge, international experience and relationships -have been identified (see Table 1). As this is a very specific industry and the number of potential producers is low, such aspects seem to affect their internationalization less. Table 3 illustrates the values of two indicators chosen to measure Molycorp Silmet's financial performance from 1999 to 2014. It can be seen that fluctuations have been remarkable. Bankruptcy prob-ability ranged between the best score 0.03 in 2013 and the worst score 0.74 in 2012, whereas most of the values remained under the critical cut-off point of 0.5, reflecting that Molycorp Silmet has been in good financial health for most of its existence. The cash flow based profitability has also fluctuated from negative to positive values, whereas the latest years indicate a positive CF/S.   Tables 4 and 5 outline the Spearman correlation coefficients between Molycorp Silmet's export market shares and export market dispersion on one hand and the profitability ratio and bankruptcy score on the other hand. It can be seen that the shares of the target market groups have only affected CF/S from period t, namely an increase in the EU's share in total sales has led to deterioration of CF/S, and on the contrary, an increase in USA's share in total sales has led to improvement of CF/S. Still, these effects are nonexistent when comparing past target market proportions (from year t) with future financial performance (from year t+1). Thus, the target market group choice has influenced cash flow based profitability only in the same year. Table  4 also indicates that with an increase in the share of the main target market group, cash flow based profitability will decrease, thus increased concentration on a single target market group can prove to be financially unviable. Again, such an association is not present in the case of lagged financial performance (see Table  5). Interestingly, none of the applied internationalization variables associate with bankruptcy probability, thus it can be proposed that this variable is more influenced by other managerial decisions, not internationalization choices. An important result is also that CF/S is positively associated with the aggregate turnover from both periods t and t+1, thus clearly indicating the presence of return to scales. Namely, larger exports will result in increased profitability.  Table 4 Spearman correlation coefficients (SCC) between two financial performance variables from period t and seven internationalization variables from period t (n=16). Thus, the following conclusions can be made from the above research. The changes in the shares of Molycorp Silmet's major target market groups affected its cash flow based profitability in the same year but were not associated with bankruptcy probability neither in the same year nor in the following year. Also, concentra-tion on a single market group reduced cash flow based profitability in the same year, but other indicators of target market dispersion did not have such an effect.

Conclusions
The results showed that several factors -mostly external -lead to Molycorp Silmet's export fluctuations. Furthermore, despite such fluctuations, the firm's fi-nancial health was mostly good. We also concluded that fluctuations in the target market shares are not interconnected with bankruptcy probability but they can increase or decrease a firm's profitability. Also, we found that lower profitability can result from high concentration on a single target market group. Thus, managers should not regard nonlinear internationalization -export volume or share fluctuations -as a sign of failure, especially as such fluctuations can often be caused by external factors. It would be interesting to study more rare metal producers in the future to find out if their nonlinear internationalization has been affected by similar factors. Moreover, their performance and its interconnections with internationalization need further attention.