How do we measure social transparency in fashion supply chains?

🕒 This report is more than 5 years old (Published Jul 25, 2017).

Historical context behind supply chain transparency issues

After Rana Plaza collapsed on April 24, 2013, in Dhaka, Bangladesh, and killed more than 1,100 workers, the fashion industry fell under widely publicized scrutiny for its negligent social practices. With consumers and non-governmental organizations aware of these issues and creating public pressure on the industry, many companies are increasingly trying to institute transparency within their supply chains to become socially sustainable.

However, supply chain transparency so far has not been clearly defined, which makes the process of evaluating social transparency difficult and often unpractical.

Particularly, after extensive research and literature review, I have discovered two conspicuous issues in this field:

1. Ambiguous definitions of supply chain transparency are coupled with (often non- standardized) methodologies of evaluation, which renders the process of evaluation dubious, and
2. More precise definitions of supply chain transparency are coupled with unclear methodologies of evaluation, which, paradoxically, makes the process of evaluation nontransparent.

While working at the MIT Responsible Supply Chain Lab, I wanted to therefore develop a methodology to measure social transparency in global supply chains. I chose the apparel industry as a case study to answer this question, given the severity of the mentioned social issues, expecting to primarily solve the issue of inconsistent evaluation methodology.

What I discovered was that law and diffusion of technological innovation were even more paramount to the global problem of compromised social sustainability. My full thesis based on this project can be found here.

My high-level approach of measuring social supply chain transparency in supply chains

My three goals in this project were:

1. Provide a clear definition of social supply chain transparency by specifying what it is and  which factors influence it through a framework;
2. Develop a methodology of quantifying social supply chain transparency; and
3. Apply it to the case of apparel industry.

Interestingly, by applying these steps, I discovered that different aspects—business, technology, and policy—should be co-optimized to improve supply chain transparency.

My final discovery was that a company cannot achieve higher transparency on its own and that collaborative efforts from the company’s stakeholders, innovation- and technology-oriented firms, as well as regulatory and legal bodies are necessary to lead the industry toward improved social sustainability.

Framework and Methodology

Supply chain transparency framework

Building on previous research, I constructed a framework that defines social supply chain transparency as the degree of external and internal disclosure, which can be evaluated in both cases via three aspects: traceability, social sustainability, and purchasing practices.

Additionally, because it is important to understand what influences the degree of disclosure, I included five “underlying influences” in the framework: product formalization and standards, third-party integration, supply chain disintermediation, supply chain communication, and the political and legal complexity of sourcing countries.

Methodology of measuring supply chain transparency

To evaluate companies and their degree of social supply chain transparency according to the above proposed framework, I developed an adaptive survey using Qualtrics software that asked respondents varying questions on their external and internal transparency, while probing for information that would give insight into the five “underlying influences.”

Because the survey was developed to be adaptive, I designed logic flows so that respondents who do not need to offer much information on their practices (i.e. barely and highly transparent brands) are not required to answer the same questions as brands of medium transparency.

The set of full questions can be obtained upon request, while the logic flows (click View full size) of the simplified questions are shown in the three diagrams below.

The survey was therefore developed to evaluate three aspects:

1. external transparency
2. internal transparency
3. “underlying influences”

To complement this analysis, I also performed content analysis and interviews, although they mostly served as backup—and not primary—source of information for the overall methodology.

Interestingly, because external transparency refers to the degree of disclosure to general public, I realized that extracting the relevant questions from the survey and answering them independently by collecting mass public data on different companies, I could perform a robust analysis of external transparency and possibly gather some insights.

The goal of such analysis was primarily to extract external transparency scores, formulate hypotheses on what might affect such scores, and draw conclusions on two “underlying influences:” political and legal complexity as well as supply chain communication.

Evaluation of External Supply Chain Transparency

60 companies were evaluated on external transparency, and were initially selectively sampled based on their business orientation (trend-oriented, neutral, or sustainability-oriented). I scored external transparency on a scale from 0 – 5 based on the following five questions extracted from the survey:

If a company has published any of the above mentioned on their website (or elsewhere that is accessible by general public), it would earn a point. Because no previous scoring system has been developed in this field at the time of the project, I used a simplistic scoring (advantages and disadvantages explained in full thesis) rubric to evaluate external transparency, in which:

1. Low transparency was valued at 0 and 1 points. 
2. Medium transparency was valued at 2 and 3 points.
3. High transparency was valued at 4 and 5 points. 

In addition to scoring each company for external transparency, I collected public data on their business orientation, size (number of  employees),  annual revenue, and location to formulate hypotheses on the correlation between these factors and external transparency scores. An abridged version of data collection (size rounded for simplicity) is shown below:

Size vs. Revenue across Business Orientation, Nationality, and Supply Chain Transparency Rank

 

Since companies from Oceania are not largely present in the dataset, it is most useful to compare US and European companies across this dataset. What is most notable is that European companies tend to dominate the space of the graph with large size and revenues (trend-oriented companies as seen in previous graph), while US companies tend to dominate the space with lower size and revenue (sustainability-oriented companies).

Finally, one can also use this scatter plot to visualize the relationship across the transparency scores: low, medium, high.

My hypothesis was that companies with a large number of employees and annual revenue would score lowest on the supply chain transparency ranking criteria, but as can be seen from the scatter plot, those companies tend to be of medium transparency.

Instead, low-transparency companies exist in the space with lower size and lower revenue. Since the criteria for ranking was based on external transparency, this could be because larger companies tend to face more public pressure to publish certain information, such as list of suppliers or their code of conduct.

Supply chain transparency ranking across business orientation and nationality

I also looked at how supply chain transparency rank compares across different criteria. This was partially visible from the previous scatter plots, but one can also visualize the relationship through simple bar charts.

The most interesting category is that of sustainability-oriented businesses: although they are the ones that constitute the majority of the high-ranking companies in terms of transparency, they are also the ones that take up a large percentage of companies with low transparency. This was an interesting finding for my thesis because it showed that, despite many efforts of sustainability-oriented businesses, many of them score low on external transparency. My hypothesis, which I later investigated and confirmed through legal research, was that they simply do not face enough public pressure from external stakeholders.

I also wanted to see if there were any interesting differences across different continents when it came to supply chain transparency.

This bar graph shows that there doesn’t seem to be any notable difference across the continents in terms of rank distribution. We can see that one of the earlier observations is confirmed: according to my ranking methodology, a larger percentage of high-ranking companies corresponds to US-based firms. This, of course, can only be extrapolated through a notably small sample in this dataset as there are very few high-ranking companies.

Publishing criteria across the three supply chain transparency ranks

It was also interesting to see how the three levels of transparency rank across the five publishing criteria.

The chart shows few interesting findings:

• In general, most brands publish their purchasing practices. Conversely, full-cost breakdown is the category with the least number of brands as only a few of them publish their full-cost breakdown. Those brands, at the time of data collection, were: Everlane and HonestBy.
• Most brands do not publish their audits either. Those brands that did so at the time of data collection were: Calvin Klein, Levi’s, Nudie Jeans, and Patagonia.
• Low-ranking brands in this dataset only publish their list of suppliers and their purchasing practices. Interestingly, none of the high-ranking brands have published their full-cost breakdown.

Statistical Analysis of Supply Chain Transparency Score

Finally, the last question was: are the three models of business orientation: sustainability-oriented, neutral, and trend-oriented truly different in terms of their transparency score? For this analysis, I wanted to implement ANOVA and detect if the difference between their mean scores was statistically significant. Given the low number of data points in this dataset and the simplicity of my scoring methodology, my hypothesis was that ANOVA would produce p-value higher than 0.05, thus signaling there is no evidence to conclude the groups are statistically different.

Given the few discrete values of transparency scores, it is clear that the boxplots also become less informative. For example, for neutral and trend-oriented companies, the median climbs up to the top of the box, and trend-oriented boxplot is missing a lower whisker, as the minimum observed value is 1 and at least 25% of the values are 1. At the same time, the boxplots show that, given the low volume of datapoints in this case, we are unlikely to observe a statistically significant difference in means between these three categories.

I also visualized this relationship through the mean-standard error graph.

Interestingly, we can see that the mean does slightly increase in terms of score as we move from sustainability-oriented to trend-oriented, indicating improvement in transparency. We can also notice a decrease in standard error as we move from sustainability-oriented to trend-oriented.

ANOVA:

##                      Df Sum Sq Mean Sq F value Pr(>F)
## Business.Orientation  2   1.03  0.5167   0.461  0.633
## Residuals            57  63.90  1.1211

What this showed, essentially, is that I could not reject the null-hypothesis, which states that the mean scores of these three groups are the same. Of course, I’d argue that we would be likely to observe a different truth in reality with a greater number of data points and a more granular scoring methodology. Furthermore, since the dataset was built through a qualitative and manual data collection, it’s possible that some of the information in the dataset is incomplete, and that the scores are indeed different. I, nonetheless, concluded there was enough difference for more qualitative, and less statistical, exploration.

Analysis of underlying influences on supply chain transparency

Qualitative legal analysis

My qualitative legal analysis first showed that international law is not an effective tool of promoting social transparency in supply chains due to decentralized and clashing nature of existing international tribunals.

A potential solution to the problem is extraterritorial jurisdiction as victims of international human rights can use it to have access to foreign courts. However, the benefits of this legal tool were indefinitely offset by the political climate present at the time of the project. Particularly:

1. The US Alien Tort Statute (ATS), which served as an exemplary case of extraterritorial jurisdiction,  got overturned by the Supreme Court in 2013.
2. The EU can mend the gaps left by overturned ATS, but its complex legal system serves as hindrance.

Therefore, pre-emptive protection of international human rights from malpractices of transnational corporations has been rendered ineffective.

Quantitative legal analysis

My quantitative legal analysis focused on establishing a back-of-the envelope correlation between disclosure of Supplier CoC, as well as purchasing practices, and compliance with the California Transparency in Supply Chains Act (USA) or the Modern Slavery Act 2015 (UK).

My expectation was to see a weaker link between disclosure of these external transparency criteria and compliance with the Modern Slavery Act due to its vague and lenient provisions. This was confirmed by the analysis.

• H1: Brands that publicly comply with the California Act will have a higher rate of disclosure of purchasing practices statements and Supplier Code of Conduct.
• R1: Correct. Results show that the California Act has an influence on external transparency scores for those brands that need to comply with the Act. Small brands (less than $100M annual revenues) are not required to comply with the Act and disclose such information.

What this means is that legal acts can have an important impact on supply chain transparency. In other words,  it is not solely the company’s responsibility to be transparent. Regulatory and legal bodies need to provide an institutional foundation for increasing the degree of transparency in global supply chains.

 Qualitative supply chain communication analysis

Regarding supply chain communication, I observed that auditing is a conspicuous issue in global supply chains. Particularly, there is a significant degree of information asymmetry between supply chain players as they share results of supplier monitoring. The table shows that there are three levels of information asymmetry in supply chains.

Likewise, another problem is that the auditing system is not standardized, which makes cross-comparisons difficult and renders the auditing process largely subjective. This, in turn, has a negative impact on both external and internal transparency.

Conclusions and Suggestions

Throughout this project, I discovered that:

1.  National regulatory and legal bodies should revise supply chain-centered acts and accords to have a more encompassing definition of transparency. Once this is implemented, national courts should institute extraterritorial jurisdiction on transnational corporations. These changes would increase the degree of social transparency in global supply chains.
2.  Diffusion of technological innovations is essential for increasing social supply chain transparency. Promising results can be achieved through digitization of supplier auditing, which would make the overall monitoring process standardized and reduce information asymmetry between supply chain players.
3.  Other professionals will hopefully be able to continue this work and deploy the adaptive survey to perform in-depth interviews and complete assessment of internal transparency. This would finalize the results and offer a well-rounded, informative understanding of potential avenues to a socially responsible future.

Note: For simplicity and clarity, only specific details and aspects of this project are explained here. The statistical analysis can be accessed in R Markdown on my github.