Revealed in Chaps. 2–4 is the prominent role of social capital, particularly networking, for MSMEs to improve productivity. In Chap. 5, we explored the role of, and the issues surrounding, the financing gap and digitalization, in which it was shown that overall Indonesia’s MSMEs continue to face financing gap problems, and their adoption of digital technology is still limited. This is despite the fact that access to finance and digitalization could significantly affect MSME productivity. What remains to be investigated is the relative contribution or strength of financing and digitalization in the dynamics of MSMEs’ productivity.

It is commonly argued that size matters. That is, compared to smaller firms, bigger firms have all the advantages they can enjoy to make improvements in productivity. Does this mean size is the deciding factor for productivity? Or, are there other factors more important than size? In the context of our survey finding, what about the role of network? How big is its relative contribution to productivity?

Cultural factor could affect the way inputs influence outputs. In our case, it is the role of culture that jointly shape the relations between size, network, and other factors with productivity. To investigate quantitatively the relative contribution of the above factors, local environment and culture needs to be accounted for as they could play some role in influencing the way those factors determine productivity. Given the fact that Indonesia is a multicultural nation with significant ethnic, religious and linguistic diversity that could influence her development process and how millions of MSMEs throughout the country operate and perform, it is even more important to include these culture-related elements in the analysis (Azis, 2019, 2020).

Fig. 6.1
figure 1

Organization of this chapter

Full size image

Since the majority of MSMEs consider network to be the most important component of social capital, our next task is to test the primacy of network along with other relevant variables including firm size and cultural factor in influencing the MSME productivity. For this purpose, we used hybrid (secondary and primary) data. Our particular interest is to compare quantitatively the extent of the contribution of network and that of firm size in affecting MSME productivity by including a cultural variable in the equation.

The bulk of this Chapter is devoted to the above task, the organization of which is depicted in Fig. 6.1. We begin in the next Section with a simple association test, using simple heat maps before applying Chi-square tests. In Sect. 6.2, we focus on comparing the role and contribution of firm size and network in affecting productivity by running the instrumental variable (IV) regression. In the model, we included culture-related variables, one as the instrument and the other as a control variable. In the set of control variables, we also included the two factors discussed in Chap. 5, finance and digitalization.

6.1 Simple Association

To the extent the results from the survey and the use of alternative approaches point consistently to the importance of network to establish linkages and interactions among MSMEs as well as between MSMEs and other stakeholders, leveraging network appears to be the most important measure that would have a profound effect on MSME productivity. Through a series of interviews conducted outside the formal survey, we heard repeated complains about a lack of access and the desire to have better interactions with other MSMEs, government apparatus, suppliers, and customers including bigger firms and potential lenders. Some were so confident that it would not take too long to make productivity improvements once a network is established. On the other hand, a number of studies have revealed that the size of business matters the most in determining productivity. Compared to what small firms can do, bigger firms with all their advantages are generally more able to conduct activities that can boost productivity, i.e., raising output and/or economizing inputs.

Analytically, we could simply test between size and network and determine which one has a bigger influence on productivity. The effect of size is probably easier to detect, but from the perspective of policy intervention it makes more sense to think of finding a possible intervention that could affect network rather than the size. The owners of the firms themselves can determine what size they can afford or prefer to have and operate. To delve into this matter, we first compare the association between size and productivity and between network and productivity by plotting data in heat maps. We subsequently run a simple association test using the Chi-square to compare the observed and expected results. The test is intended to determine if differences between observed and expected data are due to the relation between variables or just a chance.

To get the information on productivity, we gathered inputs from MSMEs’ regarding the change in their business’ productivity during the period of pre-COVID and late (or abated) COVID. After explaining the definition of productivity, i.e., value of output compared to value of inputs, two sets of questions were used: first, whether their productivity decreasing, unchanged, or increasing; second, approximate the quantity of change by using a particular range of percentage change, (0%), ± (>0–50%), and ± (51–100%). Hence, there are five alternative answers to this question.

Those who suffered from a decline in productivity during the abated-COVID period often had some excess inventories. They were more inclined not to increase production unless new orders arrived. But they still had to pay wages and rents, and also made payments to the suppliers. Most of these MSMEs operated in the non-farm sector. On the other hand, MSMEs in the farm sector felt that the COVID pandemic did not affect their productivity too much. Weather conditions and other climate change related factors produced a lot more effects on their activities. It appears that MSMEs who were able to either maintain or raise their productivity include those that used digital technology (e.g., e-commerce and social media platform), and those that diversified their products (e.g., garments industries producing masks).

For the size, we used the average annual sales as a proxy. We followed the classification based on Law No 20/2008, i.e., when the annual sales are less than IDR 300,000,000 the respondents are categorized as micro; when the sales are IDR 300,000,000–IDR 2,500,000,000 they are small; and when it is IDR 2,500,000,000–IDR 50,000,000 they are classified as medium.Footnote 1 Since our respondents were located in different regions, we deflated the sales value by the per-capita wages in each region to reflect variations in regional economic conditions.

figure a

Survey story: Traditional weavers in Boti village, Kie District, East Nusa Tenggara. They are part of the indigenous tribe that has lived for generations and is among the last Kingdom on the island of Timor (there are more than 300 Inner Boti people and around 2,500 Outer Boti). The tribe maintains their cultural values, keeps certain rituals to strengthen the human relations and social bonds and to connect humans with the sacred. Having limited contacts and network with outside parties, and with no external funding, the women weavers produce woven materials for their own clothing and few are for sales to some visitors (shown in the above picture). All the raw materials they use are natural, and they also preserve the spinning and weaving methods that are sustainable and environmentally friendly. By standard measures, they may have a relatively low productivity. But the cultural importance and the beauty of their products, along with the historical and philosophical values embedded in the motive they use, which they inherited from their ancestors, give a distinct value to the products and become the source of their comparative advantage

To get the information on network, we distinguished the quantity (number of network) from the quality part (effectiveness of network) for an obvious reason: having more networks does not guarantee gaining greatest benefits if the effectiveness (quality) of those networks is questionable. The idea is similar to the concept of “centrality” where it is not only the prominence (out-degree centrality) or influence (in-degree centrality) of connected links that matters, but also the “neighbors” connection. Thus, a network system depends not only on how many connections it has but also on how many connections its “neighbors” have, and on how many connections its “neighbors” neighbors” have, and so on. The degree of influence of members in a network also matters, and so do the strategic position and the reputation of members.

Consider the case of an MSME network in a cluster. If one MSME is well connected to another MSME who is not well connected themselves, the first MSME is “influential” (because others may depend on it) but may not gain benefits from the network. On the other hand, if the other MSME is also well connected, the first MSME can reap benefits from the network. Thus, the quality of networks matters.

We therefore constructed a “network index” by multiplying the quantity part with the quality part, leaving us with three measures of network: network quantity, network quality, and network index. Since this is neither a binary nor an easy to quantify measure, we used the following grouping to capture variations in the intensity of network: very few, few, medium, many, and a lot. We attached the following range for both the network quality and the network index: 0–20%, 21–40%, 41–60%, 61–80%, and 81–100%, respectively.Footnote 2

Fig. 6.2
figure 2

Source Own estimation

Heatmaps of network index and sales, pre-COVID.

Full size image
Fig. 6.3
figure 3

Source Own estimation

Heatmaps of network index and sales, COVID.

Full size image

From the heatmaps shown in Figs. 6.2 and 6.3 it can be seen that high productivity MSMEs (darkest red color) are associated with high network index but not with the size. On the other hand, low productivity MSMEs (light yellow) tend to be associated with lower network. No consistent association is detected between size and productivity: both the highest and lowest productivity cases are associated with stronger sales (size). Hence, while there is some degree of association between network and productivity, the association is relatively weak. In the meantime, no association can be detected between size and productivity. There is also no association between all the three variables during the COVID period (Fig. 6.3).

To have a more concrete picture, we conducted the following Chi-square analysis. We categorized the size according to Law No. 20/2008 for micro, small, and medium enterprises, adjusted by the per-capita wages, and for the productivity we used the following classification: lowest, second lowest, third lowest, fourth lowest, and highest, based upon which we fixed the percentage value as the threshold on each group, e.g., the lowest refers to those whose productivity change reached up to 20th percentile. We classified the network into low, medium, and large, and used the corresponding percentage value as the threshold, e.g., low refers to those who had a network index up to 33rd percentile. The results of the chi-square test for the two periods are displayed in Table 6.1.

Table 6.1 Results of the chi-square test
Full size table

It appears the only non-significant association is between productivity and size. This seems to defy the common view about the role of firm size. Although a network has a significant association with firm size, and it has also a significant association with productivity, by itself size does not seem to be associated with productivity. In the mean time, consistent with the arguments made in Chap. 5, the use of digital technology has a significant association with productivity. Compared to the heatmap presented earlier, the Chi-square test is more robust with respect to the distribution of the data. Although the sample size requirement is restrictive, the test does not require homoscedasticity in the data. Also, the variances among groups do not need to be equal. While instructive, however, caution needs to be exercised when interpreting the above results. In addition to a lack of causality evidence, the practical relevance and the policy implications of the results are very limited, if not none whatsoever. A more rigorous analysis is therefore needed.

6.2 Instrumental Variable (IV) Approach

As is well-known in any impact study, it is perilous to isolate a single factor as the determinant of an outcome. While having a network is critical for most MSMEs to improve productivity, other factors may also have some contributions to the improvement. In our survey discussed in Chaps. 2, 3 and 4, we had included some of those factors. They were embedded in the goals, challenges, and specific problems within the hierarchy or network. Here we conducted the test using a different approach by utilizing secondary data that include non-economic factors and the results of interviews.

It would have been desirable to apply a randomized experiment (RE) to test the role of network as part of social capital on productivity. However, given some constraints, and the limitations of RE, we instead conducted the test using the Instrumental Variables (IV) technique on the mixed secondary and survey data where problems of confounding or mixed effects are minimized.Footnote 3 Considering the importance of social capital that consists of three components, norms, trust, and network, we focused on the network variable and included on the list of explanatory variables the frequency of ethnic conflict as a proxy of a lack of trust, and the presence of indigenous communities (masyarakat adat) as a proxy for norms.

The starting point for the model framework is a proposition that size and network jointly determine the MSME productivity. However, since both are not completely exogenous as they can be influenced by other factors, we cannot run a direct regression using either of those two variables. We need to find the purely exogenous and time-invariant variables that influence the non-exogenous part but also have an effect on the relation between MSME productivity and the aforementioned two variables (size and network). This variable is to be used as the instrument in the model. The multicultural nature of Indonesia with the significant ethnic, religious, and linguistic diversity throughout many islands prompted us to use the presence of indigenous ethnic community as a candidate for the instrumental variable. This frequently overlooked factor is embedded in the way-of-life of the ethnic communities in different parts of the country whose activities include exchanges such as producing and selling products and buying goods for daily needs.

We also included a set of control variables, the effect of which may work only indirectly through the instrumental variable. As discussed in Chap. 5, the use of digital technology actively promoted by the government especially during the COVID pandemic must have had some effects on both the network and the size of MSMEs, hence it is included in the set of control variables. Another control is the size of credit outstanding that reflects a condition associated with the financing gap (also discussed in Chap. 5). MSMEs with higher credit outstanding—or facing less problems of financing gap—are likely to be those of the bigger size or having a good network.

The remaining control variables depict the socioeconomic conditions, including education level, health conditions, and ethnic conflicts in the localities/districts where the MSMEs operate. From the MSMEs’ perspective, these variables may not be seen as to have effects on their activities. Even if they are aware about the effects, it is not easy for them to comprehend the intricate link and interactions between those variables and productivity performance, let alone the mechanism behind it. It is unclear, for example, how hospital density or school density in the locality where they operate has any bearing on their business operations. Equally less obvious is the mechanism of how ethnic conflict would affect productivity (see Azis & Pratama, 2020). The complete list of variables used in the model is displayed in Table 6.2.

Two alternative measures of network are used: Network 1 is the network index capturing both the quantity and quality of network as described earlier, and Network 2 is the index that captures only the number of network (quantity). In the first stage of the 2SLS method in the IV model, we evaluate the effects of the IV and control variables on those two measures of network, and the size of MSME (listed at the top of the columns in Table 6.3). We also evaluate the regression results in which the endogenous variables are the interactions between network and size. The two types of interaction are in the last two columns of the Table. The instrumental variable (“Dummy ethnic”) is listed in the first row.

Table 6.2 List of variables in the model
Full size table
Table 6.3 Results of the first stage of the IV regression
Full size table

The results show that Network 1 gives better statistical results than Network 2, in which all control variables are significant with the expected signs. More importantly, the role of “Dummy Ethnic” is highly significant (at 5%) in explaining the variation of network index. In reality, MSMEs’ capacity to get involved in a network is highly influenced by local conditions where they operate. Though some of the conditions may not be purely exogenous—vary according to changes in other exogenous variables—the presence of ethnic group is strictly exogenous and non-time varying. This variable turns out significant and it restricts the number and the effectiveness of network (the coefficient is negative and significant).

figure b

Survey story: Interviewing a micro business owner producing a healthy drink using local ingredients in East Kalimantan. This micro business had difficulties to access affordable loans and to adopt digital technology. However, based on the owner’s explanations and clarifications, its capacity to grow depends on factors beyond financing and technology. More specifically, it is influenced by its ability to acquire reliable inputs, to meet local regulations, and to reach greater market access. Networking with suppliers, regulators, traders, buyers and other relevant stakeholders would have helped them in those areas, and could be more effective than helping them with financing

As commonly argued, ethnicity can affect interactions both within and outside the community. The rigidity of the boundaries may vary, but it nonetheless allows little or restricted interaction with the outsiders as part of their adherence to traditional values and culture. Under such circumstances, having a network with outside communities is difficult. Any intervention and influence from outside must first bridge the gap between the perspectives of the community in terms of their needs and problems and those of the outside communities. That may not be easy, as our research team have experienced during the visits to the indigenous communities of Boti in East Nusa Tenggara province and Baduy dalam (inner Baduy) in West Jawa. Extra efforts are needed to introduce new things and ideas, and it requires knowing the lens through which people in such communities perceive the world.

On the other hand, many traditions and aspects of human welfare in such communities are often more favorable than those found outside the communities, on which the latter could learn.Footnote 4 Although the small business operations inside the communities may not have an ideal level of productivity due to the absence of networking, given the numerous positive traditions and norms they have kept since their ancestors time, having a lower productivity based on a standard measure may entail only small costs to the communities and society at large. Such traditions could range from the ancient belief to preserve natural resources and environment, and to live harmoniously with others. Combined with the embedded historical merits of those traditions and norms in what they produce, the real value of the products could be much higher. The conversion of a standard indicator into a broader measure of productivity (“cultural productivity”) could bring economic benefits beyond the commercial value of the products themselves. It could bring more resources useful for the protection and applications of the indigenous peoples’ cultural value, and, if the products under consideration are protected by intellectual property laws, it could also bring more benefits that could elevate the real value of (cultural) productivity. That is, the intangible asset holdings per person (similar to the value of a product brand) are higher than the case if the intangible components are ignored.

Table 6.4 Results of the second stage IV regression
Full size table

Although the coefficient for the “Dummy Ethnic” is significant, the effect of the same set of variables on the size is generally not better than the effect on the network (Table 6.4). The only other significant variable is the number. of conflict with other ethnic groups as a share of total conflicts (“Share Conflict Ethnic”). The regression using Interaction 1 and Interaction 2 shows better results, in which two additional variables that have significant coefficients are the digital use (“Digital”) and the density of primary school measured in natural logarithm (“Primary school”).

Table 6.5 Robustness test of second stage IV regression
Full size table

The superiority of network over size is also evidenced by the results of the second stage regression, albeit milder than those in the first stage (Table 6.4). Although both network and size have a significant positive effect on productivity, the degree of significance is slightly higher for the former. To the extent the selected instruments are subjected to an exclusion test (Greene, 1997), the results of the Wald test shown at the bottom of the Table indicate that Model 1a passes the test at 10%; this is the model that we eventually use. The results of the Hausman test also show that the exogeneity cannot be rejected, albeit with a relatively low power (due partly to the low R\(^2\) in the instrumenting regressions). Note that all results of the Hausman using dummy ethnic in Table 6.4 are significant at a five percent level, suggesting that Network 1, Network 2, Size, and Interaction 1 and Interaction 2 can be treated as endogenous variables and dummy ethnic as the instrumental variable. Table 6.5 shows the results of the robustness tests using the predicted values of Network 1, Network 2, Size, Interaction 1, and Interaction 2, obtained from the first stage model. where dummy ethnic was used as the instrumental variable. Note that the coefficient for Network 1 is also positive and significant, and that of other variables have the same sign, albeit with different degrees of significance.

It is therefore clear that having a network is critical for productivity improvements. What is opined by MSMEs, as revealed by the results of our survey, is validated by the finding of the IV model based on data from the relevant variables. The many observations and anecdotal evidence about a close association between large MSMEs and their productivity appear to miss the more important channel of influence. Specifically, it overlooks the role of network. Had smaller MSMEs been able to get the same number and quality of network as the large ones, the productivity improvement of both could have been comparable. MSMEs with a more extensive network, irrespective of the size, can achieve higher productivity than those of a large size with limited network connections or meager quality. It is the network, not the size, that matters more.

Before closing this Chapter, it is useful to illustrate some concrete examples or case-based evidence showing how having a good network can bring benefits to small businesses. These examples serve to complement the key findings throughout the book. Metaphorically, if the analysis based on models and data from Chap. 2 to the current one is like a life in today’s virtual world, the case-based evidence brings that life back to reality. Observing MSMEs throughout many regions, it was clearly noticeable that those having a good network with stakeholders and performing well had a similar experience and story. In what follows are typical examples of such a story.

First is the experience of a small firm “Manika” in East Kalimantan producing handicrafts made of beads as part of the Borneo bead heritage. The products are generally used as personal ornamentation and value objects. For a number of years, the firm had performed well in terms of sales and cost efficiency, even during a crisis. Employing some 20 artisans, the owner, who inherited the business from her parents, maintained a good network with the relevant stakeholders, i.e., local government, policy makers, regulators, and other institutions. In particular, the firm has a good relationship with local offices of the Ministry of Industry, Ministry of Trade, and Ministry of Cooperative. The owner specifically indicated the case where the firm often had useful discussions with officials at the local office of the Ministry of Industry about product development and design, and potential new products the firm could develop. The local office of the Ministry assisted them not only by sharing the necessary information about inputs’ sources and market conditions, but also by providing tools such as woodcutters, sewing machines, and input materials (the raw beads). Having such regular discussions helped to ensure that the tools they received met the required specifications for producing existing and new products. Even when the officers were replaced by new ones, the firm managed to continue a good relation because the networks they built were not with the individuals, but the institution. The firm’s owners firmly believe that such networks are essential for the business to grow, to raise efficiency, and to enhance productivity. A good network was also maintained with officers in the local Ministry of Labor and Manpower and the Ministry of Tourism. This made the task of obtaining a permit and other necessary documents easier to do.

Aside from government offices, the owner also kept a good relationship with a charity agency involved in various social activities (e.g., in clean water and sanitation programs) that brought benefits to the local community. Together with BI, the agency helped to build a gallery for the firm’s products. Since, expanding the network with customers and suppliers is important for the firm to grow, having a good relationship with other institutions that are also supportive to the idea of creating a network is imperative. This is where the local BI office has been very helpful. The office assisted the firm with the promotion and participation in exhibitions, as well as in connecting the firm with new customers. As far as the network with lenders is concerned, at the time of the interview, the firm did not use any outside funding because they did not really need it. All expenses were self-funded. However, since having a good network with stakeholders is a principle that the firm holds consistently, and given the good performance of the firm, it is unlikely that the firm will have any difficulty to secure some outside funding in the future when it needs one.

Another case-based evidence is from a small agribusiness firm “Aspakusa Makmur” in Boyolali, Central Java. The main activity of this MSME is to collect and sell horticulture products, mainly vegetables, produced by farmers in the area who have been under the guidance of the local BI office for some years. Those farmers sell their products to the firm. After being sorted, weighed, wrapped, and packaged, they are then sold to the market through various retail stores. At the time of writing, the firm has five farmers as members of the group, involving more than 200 small farmers. In the beginning, these farmers sold their products mainly to the traditional and local markets, receiving low prices and with limited market. As they began to sell their products to the firm, a network was established, through which they were incentivized to improve some aspects of their production to meet the criteria set by the firm. In return, they received agreed prices and guaranteed purchases.

The firm also expanded its network. As this MSME developed a closer relationship with the local office of the Ministry of Agriculture, things had gradually improved. At one point, with the help of that office, they were approached by a technical mission from Taiwan. After a series of exchanges, the firm started to establish a network with some local chain stores. Through that network, in 2008 it managed to sell fresh vegetables in bulk for the first time to several big chain stores outside the region (i.e., Surakarta, Semarang, Surabaya, and Yogyakarta). This forced the firm to make some adjustments, for example in product quality and packaging to meet the standard set by the stores, and also in the firm’s accounting and book-keeping. In the process, the firm added a new equipment for quality control, e.g., cold storage with particular capacity and specification, a special machine with plasma ozone technology to preserve the freshness of food products.

One thing led to another. Through a contact with the regional investment board office, together with the local BI office, the firm received some help in terms of expanding the market and getting new customers. By working together with these institutions, members of the group were able to participate in activities such as exhibitions, business matching forums, and management training programs, from which they gained benefits for improving their business performance. Not only the management of the firm improved, but their sales also increased significantly. As the production and sales expanded, including for the export market, problems such as quality control and food grading became more and more important to address. Here too, a close relationship with the local office of the Ministry helped the firm to overcome several issues surrounding those problems. Also, internally the firm began to realize that as they had to start dealing with a larger market, and the number of product varieties increased (currently they managed about 80 types of vegetables/fruits including some rarely available products such as red amaranth, okra, and certain types of asparagus), it was necessary to make improvements in product sorting and packaging. Through a network with the stakeholders, the group was able to acquire new information and materials needed to support such improvements. When the COVID pandemic and the lockdown requirement hit, one of the stakeholders donated a delivery vehicle as the firm had to start delivering the products directly to the customers.

Having experienced the network-driven progress, this firm have received several awards, e.g., “Adikarya Pangan tingkat Kabupaten Boyolali 2012,” “Anugerah Produk Pertanian Berdaya Saing tingkat Nasional, kategori Produk segar berdaya saing 2014,” “Adikarya Pangan Nasional tingkat Provinsi Jawa Tengah, kategori Pelopor Ketahanan Pangan Tahun 2015.” This MSME also won a competition organized by Tim Pengendalian Inflasi Daerah, TPID (the regional inflation-control team) for its achievement in helping to reduce the regional inflation.

Other similar success stories from good-performing small businesses that had a good network with the stakeholders can be found among our respondents. Some experienced real improvements after they formed a cluster involving other MSMEs, others were able to enhance the productivity after the local bank, upon the persuasion of the local BI office, agreed to write off some of the past debt and to extend new loan. Clearly, the power of networking is conceptually proven and empirically supported by evidence.