Chapter 6: Methodological Considerations
6.4. A Mixed-Method Study of Exchange Rate Dynamics in Brazil
6.4.1. Ontological and Epistemological Background
This dissertation has adopted a Post Keynesian theoretical approach. It has argued that exchange rate dynamics in Brazil are determined by the positions and expectations of investors in short-term financial markets. These expectations, in turn, are formed under fundamental uncertainty, which makes them, and indeed exchange rate movements, necessarily context and time specific. As a result, there are no permanent causal relationships between exchange rates and fundamentals. Thus, stressing the non-ergodicity of economic processes, this dissertation has explicitly embraced the open system ontology of Post Keynesian economic thought.
196 However, this dissertation has also rejected over emphasis on the role of uncertainty and animal spirits in expectation formation. As discussed above, uncertainty
fundamentalism runs the risk of reducing economic processes to the subjective and atomistic behaviour of economic agents, devoid of social context, structures, mechanisms (Rotheim 1999). Thus, while rejecting the closed system positivist ontology and epistemology of mainstream economics which sees permanent relations between exchange rates and underlying fundamentals, this dissertation has also objected to a constructivist standpoint which would argue that exchange rate dynamics are the sole outcome of the subjective expectations and perceptions of individuals operating in foreign exchange markets. In contrast, it has inquired after the real underlying
mechanisms, processes and structures which shape economic agents‘ expectations, financial positions and thus exchange rate dynamics. In addition, it has argued that economic agents‘ behaviour shape the underlying mechanisms and structures in a process which can be described as transformative. Consequently, this dissertation‘s ontological stance could also be considered critical realist.
Given its proximity to critical realist ontology, this dissertation has pursued a
retroductive research strategy. Based on initial beliefs and hypotheses, it has generated knowledge about the underlying mechanisms and processes in an iterative empirical analysis. Two main pieces of information were taken into account in forming initial beliefs about the underlying mechanisms and structures shaping exchange rate dynamics in Brazil.
The first piece of information was observations of event regularities on the empirical level and preliminary quantitative and qualitative research; as to the former, this dissertation was motivated by the observation, described in Chapter 4, that over recent years the Brazilian Real has been subject to large swings, excess volatility and sudden and large exchange rate movements, mostly independent of underlying fundamentals. It was further observed that these movements largely coincided with short-term domestic returns and conditions on international financial markets. Preliminary data analysis, including both quantitative and qualitative data, supported these initial beliefs and
197 hypothesis.138 This preliminary analysis also pointed to the recent process of currency internationalisation.
Secondly, these empirical observations were complemented by Post Keynesian theoretical conjectures about the underlying mechanisms and structures. As discussed above, while critical realist ontology highlights the importance of deeper structures and mechanisms, their specification is up to the particular theoretical approach adopted.
This dissertation did so by focusing on one of the most important underlying mechanisms which emerges in an economy under fundamental uncertainty in the writings of Keynes: money (Arestis, Dunn et al. 1999; Rotheim 1999).
Figure 6.1 shows a schematic picture of the structured and layered reality of exchange rate determination in Brazil assumed in this dissertation.
Figure 6.1: Exchange Rate Determination from a Post Keynesian/Critical Realist Perspective
The underlying ―deeper‖ structure in the real domain is the recent process of currency internationalisation discussed in Chapters 4 and 5. This currency internationalisation has made the positions of economic agents in short-term, domestic currency denominated
138 One element of the preliminary quantitative analysis was an explorative event study, presented in Appendix 1. Initial qualitative insights were generated with interviews during the preparatory fieldwork described in Section 4.3.
Empirical Manifestations (interest rate differential; net short-term
foreign obligations etc.)
Short-term Returns Liquidity Preference Financial Structure
Yield – Liquidity Premium Uncertainty - Money Currency Internationalisation
Empirical
Actual
Real
198 financial assets an important determinant of the exchange rate in Brazil. Based on the view of the exchange rate as international money these decisions, in turn, are shaped by the domestic currency‘s yield and liquidity premium relative to the money of the system. On the actual level, these real mechanisms show themselves in the form of returns on short-term financial assets (including expected exchange rate appreciation), liquidity preference and financial structure. Financial structure also includes the institutional liquidity of a market and its foreign exchange productivity to meet
outstanding external obligations. Finally, these mechanisms can manifest themselves on the empirical level through different indicators, e.g. short-term returns on domestic assets or a country‘s net short-term foreign liabilities. Importantly, it has been argued that, given the complex, open and organic nature of economic processes and the transformative actions of economic agents, the empirical manifestations of the underlying real mechanisms will be context and time specific or might not be observable at all.
Following critical realist and Post Keynesian methodology, this layered and structured reality was analysed using mixed-method triangulation. This methodological pluralism allows ―testing‖ of the assumed mechanisms and structures in an iterative process and sheds light on the complex and structured reality assumed both in Post Keynesian and critical realist thought. Reflecting the open system ontology adopted in this dissertation, empirical results were based on a case study of the Brazilian foreign exchange market.
By employing a case-study design, a variety of data collection techniques, variables and actors can be taken into consideration, allowing for a more holistic and interpretive analysis of the hypotheses (Creswell 2003). In this vein, Mearman (2004) argues that case studies might avoid some of the problem of closure by focusing on a small group or even individual, requiring that the homogeneity required is more liked to be
achieved.
6.4.2. The Methods Applied
This dissertation has combined insights from 88 semi-structured interviews with financial market participants, with advanced time series econometrics, such as the cointegrateded VAR and multivariate GARCH models. 52 of the semi-structured interviews were conducted with currency traders, both in Brazil and in developed
199 financial markets (primarily London), which constitute the core of the qualitative study.
A discussion of the specific econometric methods will be presented in Chapters 8 and 9.
A detailed presentation of the qualitative study conducted will be subject of Section 4.3.
The qualitative study was motivated by several considerations. Firstly, given the focus of this dissertation on financial actors‘ positions and expectations, it aimed at exploring how market participants operate in short-term financial and currency markets. Based on the critical discussion of the nature of fundamentals presented in Chapters 2 and 3, emphasis was placed on interviewees‘ perceptions and understanding of fundamentals, and ultimately the mechanisms and structures underlying their trading decisions and thus exchange rate dynamics. This also included discovering any important
heterogeneity among the interviewees and the important role of institutional factors in shaping agents‘ decisions.
Secondly, the interviews aimed to explore the link between actors‘ understanding and perceptions of exchange rate fundamentals and the underlying mechanisms, on the one hand, and the empirically observed exchange rate drivers, on the other. Thus, the analytical focus was whether, and if so, when and how the underlying real mechanisms manifested themselves on the empirical level as actual exchange rate drivers.
Thirdly, the semi-structured interviews provided rich contextual information on the structure of the Brazilian and international foreign exchange markets, the different institutions operating in these markets, their motivation to do so, their client structure and main trading strategies. In this sense, the semi-structured interviews were also crucial to uncover and shed more light on the recent process of currency
internationalisation.
Results from the qualitative study were complemented with advanced time series econometrics. As discussed above, as part of a mixed-method triangulation analytical statistics can be an important complement to qualitative data to highlight different facets of the same phenomenon. In this dissertation, time series econometrics was used to further analyse the empirical manifestations of the real mechanisms and structures shaping economic agents‘ operations in the foreign exchange market, i.e. the exchange rate drivers in the context of currency internationalisation.