Financial literacy in SMEs: a bibliometric analysis and a systematic literature review of an emerging research field

Abstract

Research on financial literacy in small and medium enterprises (SMEs) has raised a significant amount of interest in recent years and has evolved both rapidly and unevenly. This paper is the first to provide a reliable, consistent, and up-to-date review of financial literacy in SMEs through the combination of a bibliometric analysis and a systematic literature review. Specifically, and after identifying the most influential agents involved in this field of research, we carried out a co-occurrence analysis of the authors’ keywords, co-citation analysis of the cited references and a subsequent in-depth analysis of a total of 88 documents published between 2005 and 2020. The findings indicate that SME financial literacy research has been primarily analysed regarding the following topics: (1) performance, (2) access to finance, (3) innovation, (4) risk attitude and entrepreneurship, (5) owners-managers, and residual contributions. The theoretical foundations that support this research structure have been (1) the resource-based view, (2) pecking order theory, agency theory and trade-off theory, (3) the entrepreneurial orientation perspective, human capital theory and upper echelon theory, and (4) planned behaviour theory. Subsequently, we developed an integrative framework on which to base proposals of important avenues for future research. Thus, this review offers a thorough and comprehensive overview of this emerging research field.

Appendix: VOSviewer

VOSviewer (https://www.vosviewer.com) is a free tool developed by Van Eck et al. (2010) that is focused on creating, viewing and exploring maps based on networks, bibliographic and text data. In this appendix, we briefly introduce the VOSviewer terminology and the steps that were followed in the construction of the map.

First, it is important to know the terminology generally used by VOSviewer. Each map is made up of a series of items/nodes (keywords, authors, organizations, sources or countries) of different weights (the size of the item denotes its importance on the map). Items can appear connected to each other through links (co-authorship, co-occurrence, citation, bibliographic coupling or co-citation), which have a strength (degree of relationship between two items depending on the bibliometric tool selected). The set of items connected by links forms a network. Items can also be grouped into clusters (based on their similarity).

Second, we can briefly discuss how VOSviewer builds the map based on a co-occurrence matrix. This process consists of three steps (Van Eck and Waltman 2014):

• Normalization As a consequence of the differences that may exist between items and therefore in their connections with other items, it is necessary to normalize the co-occurrence matrix to obtain a similarity matrix. VOSviewer, by default, uses the association strength as a normalization measure for the co-occurrence data (Van Eck et al. 2010; Van Eck and Waltman 2009):

  $$s_{ij} = c_{ij} wiwj$$

where $c_{ij}$ is the number of co-occurrences of Items i and j, and $w_{i}$ and $w_{j}$ are the total number of occurrences of Items i and j. Hence, $s_{ij}$ is the similarity between Items i and j.

• Mapping Based on the similarity matrix obtained in the previous step, VOSviewer builds a two-dimensional map using what it calls the “VOS mapping technique”. The distance between items shows their high or low relationship. To do so, VOSviewer minimizes the following function (Nees Jan Van Eck et al. 2010):

  $$V\left( {x_{1} , \ldots ,x_{n} } \right) = \mathop \sum \limits_{i < j} s_{ij} \parallel x_{i} - x_{j} \parallel^{2}$$

subject to the following restriction:

$$\frac{2}{{n\left( {n - 1} \right)}}\mathop \sum \limits_{i < j} \parallel x_{i} - x_{j} \parallel = 1$$

where n is the number of items in a network, $x_{i}$ and $x_{j}$ are the locations of Items i and j in a two-dimensional map, and $\parallel x_{i} - x_{j} \parallel$ is the Euclidean distance between them. VOSviewer makes use of a variant of the SMACOF algorithm for the minimization problem subject to the described restriction (Borg and Groenen 2005). Since the optimization problem described does not offer a single solution, it is necessary to transform it to ensure the consistency of the results (translation, rotation, and reflection) (Van Eck et al. 2010).

• Clustering Finally, to group the items into clusters, the VOSviewer clustering technique consists of maximizing the following function (Van Eck et al. 2010):

  $$V\left( {c_{1} , \ldots ,c_{n} } \right) = \mathop \sum \limits_{i < j} \delta \left( {c_{i} ,c_{j} } \right)\left( {s_{ij} - \gamma } \right)$$

where $c_{i}$ is the cluster to which Item i has been assigned, $\delta \left( {c_{i} ,c_{j} } \right)$ is a function that is equal to 1 if $c_{i} = c_{j}$ and 0 otherwise and $\gamma$ is the resolution parameter (level of detail of the clustering).