Acute proximity and the acquisition of specialised financial knowledge: evidence from sell-side equity research and the impact of the COVID-19 pandemic

ABSTRACT

The importance of acute physical proximity in the transfer of tacit knowledge is vigorously debated within economic geography. On the one hand, the characteristics of tacit knowledge favour face-to-face transfer within localised settings. On the other hand, it is argued that technological advances have reduced this territorial fix and have elevated other forms of relational proximity in importance. This paper contributes to this debate by drawing upon 70 interviews with equity research analysts, as highly specialised, tacit knowledge-intensive financial professionals. We investigate their need for acute proximity when acquiring tacit knowledge, referencing evidence provided by the COVID-19 restrictions on physical proximity. Examining lived experiences, before and during the pandemic, we find that specialised and complex knowledge is acquired by junior team members primarily through intense interactions enabled by office working. The work-from-home policies interrupted this learning mechanism and degraded their development. But as analysts gain expertise, they become more autonomous, less dependent on the office as a source of knowledge, and more reliant on temporary and remote relationships. This suggests that physical, relational and temporal proximity function as alternative mechanisms in the acquisition of knowledge and that their relative importance evolves as individual tacit knowledge is accumulated.

KEYWORDS:

• Tacit knowledge
• acute proximity
• face-to-face communication
• financial knowledge
• financial professionals
• investment banking‌
• equity research

JEL:

• D80; D83

1. INTRODUCTION

The geographies of knowledge play a central role within economic geography. The creation, transfer and assimilation of knowledge are all seen as critical components of topics as diverse as globalisation, economic development and competitiveness, agglomeration, clusters, advanced business services, world cities and financial centres. But our understanding of the various spatial dimensions of knowledge, especially in terms of how it flows across space, is constrained by methodological challenges, which often prevent the empirical validation of many theorised concepts (Beaverstock et al., 2000; Short et al., 1996).

This is particularly evident when it comes to discussions about the geographies of tacit (or implied/inferred) knowledge. The characteristics of such knowledge emphasise the role of the individual as the principal carrier of expertise, with the subsequent presumption that its transfer requires face-to-face communication, high-trust reciprocal relationships and physical proximity (Gertler, 2003; Maskell & Malmberg, 1999; Storper & Venables, 2004). Improved communication technologies, however, contest this presumption. There is a well-developed argument, also found in the management and organisational literatures, that the high-trust relationships needed for tacit knowledge transfer can now be maintained outside territorial constraints, with other forms of relational proximity assuming greater importance (Amin & Cohendet, 2004; O’Leary et al., 2014). A related and more nuanced view is that it is possible for tacit knowledge to flow between remote actors through relationships grounded in periods of temporarily manufactured proximity (Grabher, 2002; Torre, 2008).

All three perspectives have strong conceptual foundations. But the nebulous and opaque nature of tacit knowledge makes its observation and measurement complicated, especially when compared to alternative forms of codified knowledge. As such, many of the associated theories are difficult to validate, as witnessed by the debate over the need for acute proximity in its transfer. This paper addresses this issue by exploring how a subset of highly specialised knowledge intensive financial professionals – sell-side equity research analysts – acquire their individual expertise, and the relative importance of acute proximity in the process.

Equity research is one of the most knowledge-intensive functions within finance. Equity research analysts provide value to investors by using their knowledge to forecast the future performance and valuation of listed companies. The production and relative quality of their output is dependent on the accumulation of complex and highly specialised knowledge, embodied within and carried by individuals, not institutions, and tailored to specific client requirements through reciprocal relationships (Bettencourt et al., 2002). It is, therefore, typified by high human capital intensity, low capital intensity, a dependency on tacit expertise, and high individual autonomy (von Nordenflycht, 2010). Although analysts use both codified and tacit knowledge in a spiral intersection to form new knowledge (Nonaka, 2007), they are reliant on their individual tacit expertise to interpret complex and dynamic inputs, explicit and assumed, and generate future views which are codified into earnings forecasts, valuations and investment recommendations. These are modified as new information is accessed through a process of individual-specific interpretation and analysis. This analytical expertise, often referred to within the financial community as ‘judgement’ or ‘intuition’, is developed through a continual and iterative process, involving intense reciprocal interactions with colleagues, client and other contacts.

Although the factors influencing analyst expertise have been investigated within the financial literature (Bae et al., 2008; Bradley et al., 2017; Do & Zhang, 2020; Malloy, 2005), the insights that the activity offers into the geographies of knowledge and finance have been rarely explored within economic geography (Wrigley et al., 2003). This is despite the distinctiveness of equity research as a knowledge-intensive business service (KIBS) given its greater reliance on tacit knowledge to forecast future outcomes. This contrasts with other KIBS which make greater use of codified knowledge and precedent (for example, accountancy and law). This reflects von Nordenflycht’s (2010) recognition that not all KIBS are comparable in terms of the use of tacit and codified knowledge in their production processes. But it is notable that the more tacit-intensive KIBS have received less attention from geographers, in part because the individual and autonomous nature of knowledge makes them more elusive to observation, exceptions including Faulconbridge and Jones (2012), Hall (2007) and Wrigley et al. (2003).

The analysis presented in this paper draws upon semi-structured interviews with 70 participants in sell-side equity research including 45 analysts and seven research managers, primarily working at international investment banks, with the remainder from the buy-side (including analysts and portfolio managers), corporate investor relations and sell-side equity sales (see Annex).¹ Nearly all the participants were based in Asia with 30 in Hong Kong, 12 in Singapore, seven in Mumbai and five in Shanghai. The remainder were distributed across Auckland, Jakarta, Kuala Lumpur, London, Seoul, Shenzhen, Sydney and Taipei. The majority of respondents (74%) were male, although a higher percentage (38%) of the 45 sell-side analysts interviewed were female. One of the authors previously worked as the Asian Director of Research at an international investment bank and this facilitated privileged access to these highly specialised financial professionals.

The interviews were undertaken during 2021 and were designed to assess the need for acute proximity in the acquisition and accumulation of needed knowledge and information, referencing the impact of the COVID-19 restrictions on physical proximity. All participants experienced sustained periods of mandated work-from-home during 2020 and 2021 regardless of their location, often enforced by their employers who maintained restrictions often longer than official policies. There were some differences between banks in how they applied work-from-home policies in the latter years of the pandemic, but these were not significant in the earlier years. In addition, it is recognised that no research associates (the junior members of research teams) were interviewed due to difficulties in identification and access. Given the importance of individual learning within equity research, it could be argued that this limits the ability of this paper to draw conclusions. However, most of the analysts involved had supporting associates, who they were responsible for developing, and had started their careers as associates. As such, they were able to opine, with confidence, on the development trajectories of their associates, the involved learning mechanisms, and the need for acute proximity, as revealed through the COVID-19 restrictions on such proximity.

A pervasive theme to emerge from the interviews was that acute proximity remains important in the transfer and acquisition of complex tacit knowledge, particularly in the early stages of accumulating this expertise. The highly specialised knowledge required to become an effective analyst, especially the critical interpretative, analytical and forecasting capabilities, is primarily learnt within teams in a cascading process enabled by extended periods of intense person-to-person interaction and connectivity within office structures. But the loss of acute proximity disrupted this learning mechanism, with the virtual platforms enabled by communication technologies unable to replicate the intensity of interactions experienced within offices. There was, therefore, a broad consensus across interviewees that research associates had seen their development significantly disrupted by working-from-home. Not only was the rate of learning and development slowed, but the complexity of knowledge that could be transferred over virtual platforms was reduced.

The importance of acute proximity, however, is a function of an analyst’s development. As analysts accumulate expertise, they become increasingly autonomous and more mobile with the role of the office as a structure to access and create new knowledge diminished. There is a reduced need for the permanent hyper-connectivity of offices. Analysts are instead able to use their accumulated expertise to selectively identify incremental knowledge needs. These are subsequently acquired through periods of manufactured physical proximity with sources and contacts outside the office or through relationships formed in the physical but transferred to the virtual. This suggests that physical, relational and temporal proximity should not be viewed as distinct and separate propositions, but as alternative mechanisms for the acquisition of tacit knowledge which coexist and whose relative importance evolves as such knowledge is acquired and accumulated.

The following section introduces the debate on the importance of physical proximity in the transfer and acquisition of specialised and complex knowledge, especially tacit. Section 3 draws upon the interviews to outline the mechanisms through which research analysts acquire knowledge and develop their individual tacit expertise, while Section 4 provides supporting evidence on how the COVID-19 restrictions on physical proximity disrupted these mechanisms. Section 5 concludes.

2. TACIT KNOWLEDGE AND THE CONTESTED GEOGRAPHIES OF ITS TRANSFER

The geographies of knowledge play an outsized and multi-faceted role within economic geography and represent important components of the evolving literatures on globalisation, economic development, industrial clusters and other forms of agglomeration and world cities. On the one hand, the process of globalisation is seen to have been facilitated by technological improvements which have reduced the costs of codifying knowledge and its dissemination over distance (Amin & Cohendet, 2004; Baldwin, 2016; O’Brien, 1992). On the other hand, the ability to produce, modify, disseminate and assimilate knowledge is now viewed as one of the most important determinants of economic growth and development (Lundvall, 1998). Furthermore, the increasing geographic concentration of economic activity, as evidenced by the persistence of industrial clusters, despite technological advances, is seen to highlight the continued importance of spatially situated and territorially embedded knowledge (Bathelt et al., 2004; Gertler, 2003; Storper, 1997; Storper & Venables, 2004).

It is recognised that not all types of knowledge are the same with a substantial literature, building on Polanyi’s seminal work (1966, 1967), drawing a general distinction between codified (or explicit) and tacit (or implied). Although widely used by scholars, this dichotomy can be misleading since most knowledge exists on a continuum between wholly codified or wholly tacit (Howells, 2002). But at its simplest, the former can be expressed or represented in numerical or written formats (e.g., equations, data, manuals, rules, principles, etc.), stored within institutions, dislocated from its original setting, and easily transferred over distance. In contrast, the latter is inherently subjective, experiential, embodied within and carried by individuals. It is difficult to express in either written or spoken communication forms and is, therefore, inherently more local, or at least, more ‘atomistic’ given its individualistic nature (Gertler, 2003; Nonaka et al., 2000).

The two forms of knowledge are frequently presented as separate, for example, see Cowan et al.’s (2000) distinction between codified know-what and tacit know-how. At the same time, technological developments have allowed a greater proportion of knowledge to be codified, thereby seemingly diminishing the importance of the tacit variety. But this techno-centric approach potentially underestimates the interdependencies between the two forms. To be useful, codified knowledge needs to be interpreted and applied within specific circumstances, which requires tacit expertise. They should, therefore, be seen as complementary, and it is the active interaction between the two which is instrumental in the creation of new knowledge, (Amin & Cohendet, 2004; French, 2000; Nonaka, 2007; Nonaka & Takeuchi, 1995; Polanyi, 1966). After all, codified knowledge has little or no value without tacit interpretative and analytical expertise (Bathelt et al., 2004; Johnson et al., 2002). Not only is this expertise typically specific to, and carried by, individuals, but in a world of increasingly ubiquitous codified data, it could be argued that the importance of individual tacit knowledge and expertise has become more, not less, important as the differentiating factor in knowledge creation.

Within economic geography, therefore, tacit knowledge is theorised as the more valuable. It is often seen as a non-replicable competitive resource as economies become more knowledge-based (French, 2000; Howells, 2002) and a central component of the learning economy and innovation process (Gertler, 2003; Lundvall, 1998). Specifically, the interlinked literatures on agglomeration dynamics and industrial clusters both stress the importance of territorially sticky, informal, locally circulated and costly to codify tacit knowledge. This is seen as a primary factor driving the clustering of economic activities, particularly higher order. Through colocation, not only can firms access and exploit spatially limited knowledge spillovers, but also reduce the costs of acquiring knowledge (Döring & Schnellenbach, 2006; Malmberg & Maskell, 2002; Malmberg & Power, 2005). Furthermore, the clustering process stimulates the creation of additional localised and differentiated tacit knowledge specific to the cluster, which advances its competitiveness (Bathelt et al., 2004; Storper & Venables, 2004).

The broader role of knowledge spillovers in the formation of clusters is contested (Gordon & McCann, 2005; Huber, 2012), but there is more substantive debate over the degree to which geographic proximity, e.g., permanent co-location, is a pre-requisite for tacit knowledge transfer. On the one hand, it is recognised that its underlying characteristics make it ‘extremely difficult to transfer’ and require ‘considerable amounts of time and practice’ (Amin & Cohendet, 2004, p. 23). It is difficult to assimilate (Howells, 2002) and is best acquired through observation, demonstration and practice (Maskell & Malmberg, 1999). This, in turn, requires an interactive and intensive learning and acquisition process based on face-to-face communication and close physical proximity, as per Gertler’s (2003) ‘being there’, within committed, durable, stable and reciprocal relationships (Boschma, 2004, p. 2005; Storper & Venables, 2004). These tend to be found in high trust frameworks, both formal and informal, when mutual interests are defined, framed and enabled by shared languages, values, cultures and conventions (Boschma, 2005; Gertler, 2003; Maskell & Malmberg, 1999; Neely, 2018). Institutions, including firms, typically provide these frameworks. They provide the boundaries and structures within which knowledge is created, used and distributed, but also the social frameworks, contexts and behaviours, which act as the common platform for employees to interact and share knowledge, especially tacit.

The presumption that tacit knowledge is locally embedded has attracted debate with a focus on the extent to which these networks and relationships are location specific or can function over scales other than local. New technologies, for example, can replicate the advantages of physical face-to-face interaction unbound from the spatial fix and facilitate the transfer of more complex knowledge, including tacit, over any distance. This has shifted the debate on proximity from geographical to other relational forms including cognitive, institutional, organised and social (Amin & Cohendet, 2004; Bathelt & Li, 2014; Boschma, 2005; O’Leary et al., 2014; Torre & Rallet, 2005). The argument is that the factors necessary for the exchange of tacit information, including trust, shared values, common language, etc., are embedded within shared relationships, including social networks outside organisations, and these can now be maintained over distance between people who may never meet in person. This, Amin and Cohendet concluded in strident terms, makes a ‘mockery of the idea that spatial proximity and “being there” are one and the same’ (2004, p. 108). From their perspective, the transfer of knowledge, even tacit, is no longer dependent on location nor distance, but on relationships not necessarily anchored within territorial confines.

This view has been criticised. One argument is that it overstates the ability of organisational structures to facilitate and encourage the consistent flow of knowledge across locations (Morgan, 2004; Rutten & Boekema, 2012). There is also a view, evident in the management and organisational studies literature, that virtual relationships can never fully replicate those formed in the physical, especially in terms of the required connectivity and underlying trust to maximise knowledge transfers (Kolb et al., 2008). While permanent co-location facilitates and encourages the transfer of tacit knowledge through greater connectivity, intensity and a higher frequency of unstructured knowledge transfers (Boschma, 2005), increased distance conversely reduces the opportunities for informal contact (Kiesler & Cummings, 2002). But an emergent view is that even if face-to-face interaction and physical proximity remain important for the transfer of tacit knowledge, these may not necessitate permanent co-location. They can instead be achieved through periods of temporarily manufactured physical proximity within shared institutional, organisational, or cognitive frameworks (Grabher, 2002; Torre, 2008). But while temporary co-location should, at least from a conceptual basis, allow for the transfer of needed knowledge and the formation of high trust inter-personal relationships, empirical evidence to-date has not been supportive (Müller & Stewart, 2016).

The ‘tacit knowledge transfer needs physical proximity’ debate is extensive but suffers from two methodological challenges. The first is that many of the conceptual mechanisms explaining how knowledge is acquired and transferred lack empirical validation (Balland & Rigby, 2017; Beaverstock et al., 2000; Döring & Schnellenbach, 2006; Short et al., 1996). This reflects the difficulties of defining, observing and measuring the creation and circulation of knowledge, especially tacit (Cummings, 2004; Foray, 2004; Howells, 2002). As a result, attempts to empirically validate hypotheses tend to rely on the outputs of presumed knowledge processes, for example patents (Balland & Rigby, 2017; Jaffe et al., 1993; Ponds et al., 2010), rather than the generative mechanisms involved. This creates the risk of ‘circular causation’ with outputs used to ‘prove’ the existence of theorised mechanisms (Malmberg & Maskell, 2002, p. 435). This reflects the difficulties of measuring tacit knowledge but also the broader challenges geographers face when engaging with individual learning and knowledge transfer processes, instead of those at the level of firms, networks and economic systems (Malmberg & Power, 2005). This highlights Rutten and Boekema’s (2012) call for a focus on individuals as the principal learning agents within any system or network.

The second challenge is the risk of generalising any findings or hypotheses relating to knowledge transfer beyond the specific circumstances under investigation. This reflects the extreme heterogeneity embedded within the entire concept of ‘knowledge’, across its definition, creation, use and transmission. To an extent, the tacit versus codified categorisation seeks to address this complexity but it also risks simplifying the issue since the arising duality conflicts with the proposition that most knowledge can exist simultaneously in both forms (Howells, 2002). Furthermore, how knowledge is used and created differs significantly across individuals, institutions and activities. This is apparent when considering the various knowledge-intensive business services (KIBS) which, despite the generic term, are rarely comparable in terms of knowledge processes, and exist on a spectrum of knowledge intensity (von Nordenflycht, 2010). There are also frequent and substantial differences in the relative importance of tacit and codified knowledge as inputs. Activities predicated on forecasting the future, e.g., equity research, investing and strategy consultancy, are inherently more dependent on individual tacit expertise than those assessing current circumstances using codified data and precedent, e.g., accountancy and law.

Recognising these challenges, we seek to contribute to the literature by exploring the mechanisms involved in the creation, accumulation and transfer of tacit knowledge by investigating the behaviours of a specialised and knowledge-intensive subset of financial professionals: sell-side equity research analysts. They work for equity brokerages, often within investment banks, and provide value to investors (their clients) by using their individual tacit expertise to forecast the future performance of companies based on an assessment of the most probable outcomes for their future profitability and returns. This requires analysts to assess the most likely scenarios across a multitude of determining factors including revenues (which, in turn, are determined by competition, products offered, prices, end-user demand etc.), operating costs (including salaries, property and other infrastructure costs, production inputs and commodities) and capital expenditures, especially those required to support revenue growth, and upgrade and maintain existing production infrastructures. In addition, there are other less tangible aspects of a company’s outlook which may influence its future performance, and which analysts need to consider and incorporate into their assessments, including technological developments, government policies and management capabilities.

Analysts operate, therefore, in an information-intensive and time-sensitive environment with multiple inter-dependencies creating a high degree of complexity and uncertainty, compounded by significant spatial and temporal factors. To develop accurate, and hence, valuable, forecasts and recommendations requires highly developed individual skills and expertise, based on deep specialised knowledge learnt and formed through experience. The factors influencing analysts’ relative interpretative, analytical and forecasting capabilities have been extensively explored in financial economics, including industrial expertise (Bradley et al., 2017), proximity to corporates (Bae et al., 2008; Jennings et al., 2017; Malloy, 2005) and available supporting resources (Fang & Hope, 2020; Gao et al., 2021). But the same literature also provides evidence that proximity to colleagues is important. Do and Zhang (2020), for example, demonstrated that the hiring of a well-ranked analyst into a brokerage positively impacted the performance of incumbent analysts, while Groysberg and Lee (2010) found that analysts working with higher-ranked colleagues were less likely to move to a competitor. Both these are potentially indicative of the importance of tacit knowledge transfers within equity research teams.

The judgement, expertise and capabilities offered by any analyst, and his/her immediate team, is the cumulative function of his/her acquired tacit knowledge, as reflected in interpretative and analytical skills and access to other source of knowledge and information, including colleagues, clients, regulators and industry contacts. This creates a degree of path dependency with successful, or unsuccessful, past behaviours being passed through team hierarchies. Although this is comparable with other KIBS (Gertler, 2001), equity research is much more dependent on tacit knowledge and individual expertise given its forward-looking nature (more comparable, for example, to management consultancy than accountancy or law). While analysts use codified knowledge, e.g., standardised valuation methodologies, the all-important inputs to these frameworks are individually tailored to specific, often highly dynamic circumstances. To produce accurate forecasts and value-creating recommendations, analysts need advanced expertise acquired and developed through the accumulation and modification of tacit knowledge. This rest of this paper, therefore, investigates how analysts acquire this knowledge and more specifically, the importance of acute proximity in this process, with reference to the impact of the enforced work-from-home policies implemented during the COVID pandemic as a unique natural experiment.

3. HOW ANALYSTS ACQUIRE KNOWLEDGE AND LEARN THE RESEARCH TRADE

Equity research is a highly autonomous activity with the relative value and competitiveness of individual analysts determined by their ability to use acquired knowledge and skills to generate accurate forecasts and recommendations. This expertise is highly individual-specific and is acquired and modified over time through iterative and intense learning and development processes.

There are two broad development paths for analysts to achieve a minimum level of required expertise. One is to acquire the specialised knowledge in a related organisation, e.g., management consultancy or corporate, before carrying it across into equity research. Upon joining equity research, they can immediately apply this expertise and become a writing analyst covering companies. This was the route one of the authors followed when he moved to equity research from strategy consultancy. This path is, however, less common than the second option which involves joining the function in a junior role, often as a university graduate, and working as a research associate over a multi-year period under the tutelage of an experienced analyst. As they develop their expertise, associates will either be promoted or leave, with the better performers being given coverage responsibilities within four-to-six years of starting. Most equity analysts come through this route although they may move between different research houses during their development.

Research associates primarily acquire knowledge through two internal mechanisms. The first are formal and structured development programmes which some, although not all, research houses offer. Undertaken in small group sessions, typically involving a few hours every week, these provide codified frameworks and general guidance across a range of skills and capabilities considered important within research, including financial modelling, valuation methodologies and communication, both written and spoken. This is often considered necessary as many associates arrive with limited applicable technical expertise, even when their university education was in aligned subjects. A senior analyst at a major American bank, for example, complained how one of his associates had an engineering degree and an MBA from leading universities but could not build a ‘simple discounted cashflow valuation model’ (IP04). Another analyst detailed how on her first day, she was given a valuation model and told to update it which was a problem as she ‘had never used Excel before in such a way and had no idea where to start’ (IP50), while a research manager expressed his surprise at how limited the applicable skills of many graduate and MBA hires frequently proved to be (IP22).

The effectiveness of such structured training is, however, debated. On the one hand, guidance on how to write reports, present analyses and format models is viewed as valuable in providing a foundation for subsequent development. It is also a core component of a process to replicate and embed previously successful behaviours across different teams. This is particularly important in teaching the ‘house presentation style’ and maximising consistency across analyst outputs and behaviours. One investor, for example, complained that the inconsistent formats of sell-side valuation models meant that he often had to spend valuable time working out the structure and flow of models produced by different analysts, even from the same house (IP62). Such training could, therefore, impose standard (codified) model templates, provide best-practice examples of how to write, and thereby, improve the quality of the research product.

However, it is recognised that such programmes could not provide nor teach the interpretative skills necessary to assess the future implications and potential outcomes of current circumstances and new information, across a wide range of very different industries. They could not teach, for example, what assumptions and estimates should be applied to accurately forecast a company’s earnings or define the most probable base-case valuation, even though it is this capability which clients value above all else. As one analyst put it, ‘we are being paid to predict the future, not to report the past’ (IP40) or as another explained, ‘what matters is not the what but the so-what’ (IP26). In other words, it is not the relatively ubiquitous codified knowledge that clients value, but the specialised tacit expertise carried by individual analysts, as reflected in their forecasting and predictive capabilities. In fact, the ability to accurately forecast revenues, margins and costs is considered so valuable within finance that it is often referred to in hallowed terms as ‘intuition’ or ‘judgement’ and analysts with such skills are prized by clients, echoing Clark’s (2018, p. 273) suggestion ‘that there is a premium on individual skills and expertise in financial markets’.

The respondents generally thought that these skills cannot be taught in a structured manner as they are often situationally specific and require application of logic and expertise to a unique and frequently changing set of circumstances. Every industry has different earnings and valuation drivers, and many analysts define their expertise not by reference to their technical valuation or communication capabilities, but by their ability to define, understand and forecast the earnings and valuation drivers of specific sectors. They are not generic ‘equity analysts’ but ‘industrial’, ‘consumer’ or ‘technology’ analysts. Even these categorisations do not reflect the degree of individual specialisation. Consumer analysts can be ‘luxury’, ‘staples’ or ‘discretionary’ focused, while technology analysts frame their expertise by reference to ‘hardware’, ‘memory’ or ‘semiconductors’.

It was widely recognised that such highly specialised capabilities and knowledge are acquired and developed through a knowledge cascading process enabled by sustained periods of observation, practice, experience and assimilation alongside more senior colleagues. This is the second mechanism through which associates learn and create knowledge. There was as much emphasis throughout the interviews on learning-by-watching and learning-by-listening as learning-by-doing, with tuition, guidance and examples provided through close contact and engagement with senior analysts. Through observation, associates learn and repeat successful behaviours, and incrementally adapt them to suit their own strengths and weaknesses. To an extent, this reflects the informal management processes adopted within firms typified by high human capital intensity and autonomous working practices (von Nordenflycht, 2010). Given the required specialised knowledge is embodied within individuals and not institutions, it can only be transferred on a person-to-person basis rather than codified, retained and transferred within institutions. As such, the knowledge transfer process is intense and highly intimate. As one senior analyst explained, juniors ‘learn by watching you and working closely with you’ (IP04), while another recounted the maxim repeated by her senior analyst when she was starting her career: ‘just sit, don’t talk, listen and watch everything I do’ (IP37).

This process is usually facilitated by analysts sitting in acute proximity to their supporting associates, with teams grouped together within open-plan offices. Not all respondents liked such office formats since it impeded their ability to call clients or to secure the necessary absence of distractions to think. But there was widespread agreement that they did encourage a greater degree of intra- and inter-team communication, something which managers prioritised, reflecting the scholarship on the role of offices as a structure to maximise learning (Duffy, 1997; Thrift, 2000). A director of research, for example, detailed how, based on his own experience, he had moved his senior analysts out of their glass-fronted offices into an open-plan seating format. As he explained, ‘I was watching senior analysts keep getting up, walk out to their juniors, have a short conversation, and then walk back to their offices. And they would be doing this all the time. I concluded it would be just easier for all involved to be next to one another’ (IP22).

Such acute proximity results in a working environment typified by frequent, spontaneous, disjointed, informal and highly iterative interactions; an even more intense version of Storper and Venables’s (2004) ‘buzz’ (a term used unprompted by some participants to describe the office environment). Issues and questions are raised and resolved in rapid succession. One analyst detailed how she learnt by her senior sitting beside her and frequently telling her, ‘No, that is not the right way, do it this way’ which resulted in her ‘learning quickly what to do or not to do … and it made me a better analyst’ (IP44). Another explained how he would have models open on his computer with the junior sitting next to him and would change forecasts while explaining the changes and answering any questions his junior may have (IP14). Similarly, another outlined how his juniors would spend their first few months with him sitting alongside him and watching him build models, write notes and talk to clients, with him answering their questions in real-time (IP32). Open-plan offices also meant that associates could learn incrementally by over-hearing conversations within other teams. This was likened to a process of osmosis whereby a variety of discussions are continually overheard and each, in turn, contributes an incremental idea or new information to the associate’s overall understanding (IP56).

Through this intense interaction, undertaken over multiple years, associates acquire and accumulate interpretative and forecasting capabilities specific to a certain industry. At the same time, trust is formed as associates develop relevant expertise and become more valued by the respective analyst, often within highly personal bonds. A Hong Kong-based analyst joked about how he spent more time with his supporting associates than with his family and often knew them better than his own children (IP04). Another recognised that the intensity of the relationship meant that boundaries frequently became blurred with a near complete integration of working and personal lives (IP17). One healthcare analyst, for example, dined with his associate every night before returning to the office to carry on working (IP25), while another bought his associates a coffee-making machine and would take them out once a month to ‘thank them for their work’ (IP51).

This knowledge transfer process is resource intensive and imposes significant costs on senior analysts, and not all embrace the burden of knowledge transfer with comparable enthusiasm. This was despite an overall recognition that the short-term costs of developing their associates are more than offset by the long-term benefits. As one research manager argued, ‘it really is a case of short-term pain, long-term gain. Those analysts who spend less time developing their associates suffer more as a result’ (IP28). Another explained, ‘I keep telling analysts that the point to understand is that this matters for you selfishly because if your associate is a better associate, you profit because … you don’t do as much work and that is good for you’ (IP11). Not that this process is always successful. The interviews provided several examples of failed associate-analyst relationships, usually because analysts were either unwilling to commit the time necessary to develop the associate or because they concluded that the associate was unable to achieve the required standards. But the optimal outcome is for associates to achieve a certain level of expertise not just so that they can more effectively support the analyst, but also ensure a certain level of duplicated knowledge within the team.

This acquisition of expertise and the subsequent duplication of knowledge by associates has a direct impact on the mobility and autonomy of more senior analysts. There is a natural developmental process whereby analysts become more autonomous, less dependent on office functions, and more mobile as they acquire individual expertise. To an extent, this process is forced. In order to access new and differentiated information and knowledge, analysts need to access external sources (e.g., clients and corporates) on a face-to-face basis through short periods of temporary proximity (Bratton & Wójcik, 2022). These trips can involve weeks ‘on-the-road’, often in different time-zones, with some analysts spending more than half their working time outside the office. However, analysts are expected to maintain active coverage and such extended periods out-of-the-office are only possible with the in-office capabilities provided by supporting associates. A well-trained associate can support a travelling analyst by drafting reports, updating models and communicating ideas to, and answering questions from, both internal colleagues and external clients. The mobility of analysts is, therefore, effectively anchored in the office by their associates and the greater the associate expertise, the less office-bound analysts become.

There is another process whereby senior analysts voluntarily disengage from the office once they have accumulated a certain level of knowledge and developed strong high-trust relationships with internal colleagues. Experienced analysts frequently viewed office-working through a different lens than less experienced colleagues. Although many senior participants accepted that office-working had benefits, especially in terms of creating a structure with clearly delineated time boundaries and working patterns, there was less emphasis on the office as a place to acquire knowledge, especially tacit. Physical proximity is still viewed as important, but the sources of new knowledge increasingly move outside the office and are accessed through short periods of manufactured proximity. Such temporal proximity is highly effective for more experienced analysts because the parties involved have sufficient expertise to know what specific knowledge is to be transferred or acquired. As a result, there is less urgency across more experienced analysts to be in persistent acute proximity to colleagues since they gradually become less valuable as a new knowledge source. One senior analyst ‘needed to be in the office to do certain things, but not as much’ (IP04), while another went as far as to argue that being in the office was ‘a distraction and does not really help me’ (IP51). As an extreme example, a head of research recounted how one of his best performing analysts was ‘never in the office and his desk never used. And yet he was consistently one of [our] most successful vote getters’ (IP14).

The above findings, therefore, suggest a distinction in the need for the acute and persistent physical proximity enabled by offices when acquiring knowledge according to the relative accumulated expertise of the individual. Whereas research associates are highly dependent on the hyper-connectivity and intense interactivity of office structures to acquire and accumulate the knowledge required for their own development, more experienced research analysts often view the office ‘buzz’ as a distraction and internal connections as less important. Instead, their more valuable knowledge sources typically transition to outside the office’s physical boundaries to be accessed on a temporal basis within high-trust relationships. This distinction became more apparent during the COVID-19 pandemic, as detailed in the following section.

4. EVIDENCE FROM THE UNEQUAL IMPACT OF THE COVID-19 WORK-FROM-HOME POLICIES

The above distinction on the need for acute physical proximity in the acquisition of tacit knowledge is supported by findings from the interviews that the effects of the COVID-19 restrictions on physical proximity, including the work-from-home policies, were not experienced uniformly across analysts and their associates.

Although some writing analysts experienced initial difficulties with the transition from office to home (for example, reduced home computer processing power, lack of screens and printers, and poor broadband connectivity), many welcomed the opportunity to work-from-home in large part because they felt released from the confines of the office and its intense working environment – from what Kolb et al. (2008) termed ‘hyper-connectivity’ – as well as the need to travel. This gave them substantial additional time to develop new understandings and insights. Many analysts, for example, reported that the quality of their written product improved during the periods of lockdown, although it was noted that the increase in product flow was not commensurate with the increase in hours dedicated to writing, i.e., productivity declined. At the same time, more experienced analysts were able to carry relationships established over many years through physical interaction to the virtual, both with internal colleagues and external clients. As such, they were still able to access their information and knowledge sources, at least initially.

The general view across respondents, however, was that the intense, disjointed, spontaneous and often micro-level interactions facilitated by office structures proved difficult to replicate virtually, to the detriment of the more junior team members. Despite access to technologies enabling remote team-working, e.g., video-conference platforms and shared drives, it was recognised that these created a degree of friction and necessitated additional effort. As one analyst explained, ‘theoretically I can set up MS Teams and have my model visible to everyone else, but all that adds another hour or two and it is just faster to do it yourself. There is a reason why things move faster when you are together’ (IP13). This sense that time is scarce and that anything which extends or complicates a process is inherently disruptive, was pervasive throughout the interviews. Another participant outlined how in the office, he would conference his associate into client calls simply by putting the client on speaker, a process which according to him, involved ‘milliseconds’. Outside the office, however, he found conferencing in his associate to calls ‘very problematic’ as clients are ‘impatient’ and do not want to be put on hold during the call set-up, even for a few minutes (IP18). In addition, some analysts simply did not have the capabilities at home to use the various platforms. An India-based analyst noted that the village he had retreated to during the pandemic, lacked the broadband bandwidth to support video platforms. As a result, he became reliant on a weak mobile signal which led to ‘difficult and frustrating calls with clients and colleagues’ (IP54).

Given resource pressures, especially time, this resulted in a weakening of previously strong internal working relationships, including those between analysts and their associates. As one Sydney-based senior analyst outlined, the intense office interactions he previously had with his associates were ‘just not happening in this kind of environment. It just takes too long to send models back and forth. It is just easier to say, ‘stuff it, I will do it, I will fix it, answer it and get it done’ (IP06). Similarly, an analyst in Hong Kong complained that team relations were ‘not as tight as we were’ with substantially reduced interactions and a decline in team morale (IP08), while another in the same city detailed how she would regularly conference her associates into calls with clients and corporates on an ad-hoc basis when in the office, but this proved difficult when working-from-home. As a result, the nature of her associate interactions had changed markedly from frequent and unplanned, to occasional and scheduled (IP36).

Another implication was that the loss of inter-personal connectivity previously encouraged by offices often reduced the pressure analysts felt to engage with other colleagues, including their associates. Sometimes this was a deliberate action with the functionalities of communication technologies being used to create barriers to engagement. As one Hong Kong analyst commented, ‘in the office, associates can pester you or make sure you do not leave for home without answering their question. When at home, it is much easier for me to ignore and deflect’ (IP34). Another outlined how she would prevent interruptions when working at home by muting or blocking incoming calls, even from colleagues (IP36). Sometimes the reduced sense of needing to engage with colleagues was simply a function of ‘out of sight, out of mind’ (as per Kolb et al.’s ‘hypo-connectivity’ (2008, p. 183)). The lack of acute proximity meant that interactions moved from being spontaneous and informal to being formal, scheduled and often forced. As a Mumbai-based analyst with a large supporting associate base noted, ‘it is much harder to maintain relationships [with my associates] outside the office especially as I do not have the time to call each of them for 15-to-20 min every few hours’ (IP11). Behaviours evolved to reflect the new circumstances and, subsequently, the nature and frequency of interactions changed. One analyst in Hong Kong noted that he would regularly go for coffees with his associate when in the office but had gone for days without speaking to her during the lockdowns, which made him feel ‘guilty’ for not providing the right guidance and tuition (IP47).

This erosion of inter-team relationships had a punitive impact on associate development. It was widely accepted that the transfer of specialised knowledge and expertise from senior analysts to junior team members, previously facilitated by acute proximity, was interrupted by the enforced physical separation, especially as communication technologies proved unable to fully replicate the intensity of the office environment. As one research manager asked, ‘the real question is, are they really learning at home?’ (IP13). And the usual answer was ‘no’, at least not to the same extent when in the office, with many analysts expressing concerns about the development of their associates during the pandemic. One senior analyst in Mumbai concluded, ‘for my associates, their learning goes down significantly at home because sitting next to me they learn a lot but staying at home it’s very tough for them to do so. So in my view the learnings of my associates have gone down certainly in the last one year’ (IP07). A Sydney based head of research expressed a similar concern. He was worried that ‘the juniors are not learning anything really. Their level of development has dropped off significantly … their career development is stalling’ (IP13). This was a shared view. Another research manager concluded that his associates had ‘effectively lost a year of training and development’ during 2021 (IP22), while a Singapore based analyst detailed how his company’s return-to-office plan prioritised the junior team members as ‘their need to get back into the office is significantly greater’ (IP62).

Some research managers were not surprised by the slowdown in junior development in response to the lost acute proximity since they already had reference to precedent. In response to budget pressures, many houses now use offshore associates, i.e., supporting associates located in a lower-cost country. They typically undertake lower-order tasks such as maintaining datasheets and updating models after earnings. Although offshore associates are within the same organisation, with relational proximity, they are not viewed as comparable to those co-located with their seniors, with many analysts struggling to develop their offshore associates beyond a certain level of expertise. The difficulties of tacit knowledge transfer over distance mean that the value of offshore associates is considered lower than their co-located peers. Some heads of research, for example, noted internal assessments indicating that offshore associates are typically only a quarter to a third as effective as their co-located peers, a loss accepted on cost grounds (IP22, IP24). And if an offshore associate does show promise, they are often relocated to their same office as their analyst for development purposes; again demonstrating that acute proximity is necessary to acquire more advanced and complex expertise.

5. CONCLUSIONS

The importance of physical proximity, especially permanent co-location, in the transfer of tacit knowledge is one of the more contested debates within economic geography, primarily due to the inherent difficulties associated with validating many of the advanced theories and concepts (Balland & Rigby, 2017; Beaverstock et al., 2000; Short et al., 1996). This study, however, provides new empirical insights into how tacit knowledge and expertise is acquired by a subset of highly specialised and knowledge-intensive financial professionals. Specifically, our findings provide evidence that acute proximity remains valuable for the transfer and accumulation of tacit knowledge, as per Bathelt et al. (2004), Gertler (2003), Morgan (2004) and Storper and Venables (2004). This is revealed through two mechanisms.

First, we demonstrate that acute proximity remains essential in the development of junior team members. Research associates typically enter the industry with limited applicable expertise and, as a result, embark on a rapid learning and development trajectory. This early-stage acquisition of specialised and complex knowledge by research associates is primarily through intense, intimate, informal, spontaneous and repetitive face-to-face interactions with senior analysts through which knowledge is cascaded within team hierarchies. This is a multi-year process, involving learning-by-watching, learning-by-listening and learning-by-doing, through which associates acquire and develop the needed interpretative, analytical and forecasting expertise to become a writing analyst. This process is enabled by the acute proximity, hyper-connectivity and intensity of interactions experienced within office structures.

Second, we provide evidence that the COVID-19 work-from-home policies disrupted this learning mechanism, with virtual platforms unable to replicate the ‘buzz’ and intensity of office environments. The challenges of transferring tacit knowledge remotely over virtual platforms were reflected in the widespread views that offshore associates are persistently less effective than their co-located peers and that co-located associates had ‘lost’ significant learning time when working-from-home. This was a view widely expressed by analysts regardless of location. The views expressed by participants in India, for example, were consistent with those in Australia, China, Hong Kong and Singapore. And although the surveyed sample was primarily Asia-based, follow-up conversations with analysts in Europe and the United States have highlighted comparable themes. In addition, this loss of associate development time was reflected by managers prioritising their return to the office when permitted and, although not tested in the interviews, it is presumed that they were able to return to their pre-pandemic development trajectories once back in offices.

But the study also suggests that once analysts acquire a certain level of expertise, the office becomes less important as a structure to facilitate such knowledge acquisition. Analysts become increasingly autonomous with experience and consequently, less dependent on permanent and high-frequency connectivity with colleagues to acquire knowledge. Their connectivity requirements evolve with the reduced need for the hyper-connectivity within offices (Kolb et al., 2008). In fact, senior analysts often consider the ‘buzz’ and ‘noise’ embedded within office environments as an impediment to their knowledge production. They instead use their accumulated expertise to selectively identify incremental knowledge needs, which are subsequently acquired through temporary periods of manufactured face-to-face proximity outside the office or through relationships formed in the physical and subsequently transferred to the virtual. This supports the arguments of Amin and Cohendet (2004), Grabher (2002) and O’Leary et al. (2014), although this dislocation is only possible because of the expertise carried by experienced analysts and acquired earlier through acute proximity.

Reflecting von Nordenflycht (2010), we recognise that not all knowledge-intensive business services are comparable and that equity research, given its forward-looking nature, is relatively unique in being significantly more reliant on individual tacit knowledge as an input to its production than other KIBS, which are generally more dependent on precedent and codified knowledge. But the tacit-intensive nature of equity research means that it provides insights applicable to the broader debate on the need for physical proximity in the acquisition and transfer of tacit knowledge. This study, for example, confirms the continued need for physical proximity in the transfer of tacit knowledge, especially when individuals are in the early stages of their knowledge development. It also highlights the continued importance of the office as a mechanism to enable the intense, intimate, unstructured and spontaneous interactions necessary for this transfer. Once analysts accumulate a level of tacit expertise, however, other forms of proximity assume greater importance when acquiring new and incremental knowledge. There is a weakening of the spatial fix with more experienced analysts less ‘anchored’ to office structures than their more junior colleagues, at least in terms of knowledge acquisition. As such, the highly individualised nature of equity research expertise already provides more experienced analysts with the autonomy, mobility and control that many other financial professionals now aspire to (Cockayne & Treleaven, 2023). There was, for example, a tendency for senior analysts to work outside office confines even prior to the COVID-19 pandemic.

This research, therefore, reveals the nuances involved in the ‘tacit knowledge transfer needs physical proximity’ debate, reflecting Faulconbridge’s (2006, p. 517) call to be ‘more subtle in our arguments about its geographies’. This study suggests that physical, relational and temporal proximity should not be viewed as distinct and separate propositions, but as alternative mechanisms for the acquisition of tacit knowledge which coexist and whose relative importance evolves as knowledge is acquired and accumulated. The need for acute proximity and co-location with colleagues within office structures is critical in the initial accumulative stages and without the intense face-to-face interactions enabled by office-working, the complexity and sophistication of knowledge that can be transferred is limited. This highlights the continued learning function of offices (Duffy, 1997; Thrift, 2000). At this stage, other forms of relational proximity have more marginal influences, as seen in the difficulties of transferring knowledge to offshore associates within the same organisation. But once an individual has accumulated sufficient tacit expertise, other mechanisms, sources and forms of proximity, including temporary and relational, assume greater importance when acquiring incremental specialised information and knowledge Bratton and Wójcik (2022).

The analysis in this paper presents various potential research directions. We recognise, for example, that finance remains highly gendered (Pollard, 2013) and that perceptions on the role of the office as a structure for learning, development and knowledge transfer may differ between male and female analysts, especially if the reliance on acute proximity as a development mechanism fosters and entrenches patrimonialism (Neely, 2018). While we did not identify any significant nor consistent gender differences during the interviews (and also note that finance in Asia tends to be more gender-diverse than in Europe or the US), this is a question that merits further investigation. Related is that this paper did not explore the impact of work-from-home on organisational power structures and networks (as highlighted by Cockayne & Treleaven, 2023). It was notable that some research managers made passing references to the importance of the office as a command-and-control mechanism when justifying why they were keen for their teams to return to office working, although this topic was not fully explored in the interviews. Similarly, some of the interviews suggested that offices serve an important social function for junior team members, especially in cities such as Hong Kong where living space is limited, while more experienced analysts frequently highlighted the role of the office as providing a defined working structure and environment. These are all potential avenues for further research and analysis.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

Notes

1 The interviews were undertaken with the approval of, and within the ethics framework defined by, the University of Oxford’s Central University Research Ethics Committee (CUREC reference: SOGE 1A2020-9). All involved respondents provided consent to be involved and for their views to be documented at the beginning of the respective interview(s). All participation was voluntary, no personal information was collected, confidentiality was assured, and any supporting quotes used in this paper are on an anonymous basis.