Is there a global convergence of management foci in city credit quality assessments? A computational analysis approach

ABSTRACT

Financialization of urban governance is a process taking place with varying degrees of depth around the world. A key intermediary and a product that many cities must face to gain access to capital are credit rating firms and their opinions of underlying credit quality. Given the required economies of scale and scope with information intermediation, three global firms have emerged to offer methodologies for fundamental components of subnational government risk, which they claim follows a standardized credit assessment schedule. Computational text analysis offers insights about a crucial credit quality assessment mechanism as well as the potential convergence in management foci of city credit quality assessments globally. Specifically, the findings will assist in unpacking the management factors of city credit quality, with implications for both cities and investors in an increasingly financialized environment of urban governance.

KEYWORDS:

• City credit rating
• management foci
• credit rating convergence
• fiscal governance

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here

Notes

1. Each credit rating firm has an International Public Finance division that specializes in regional and local governments around the world–Fitch Ratings (https://www.fitchratings.com/site/ipf), Moody’s Analytics (https://www.moodys.com/researchandratings), and Standard & Poor’s Global Ratings (https://www.spratings.com/en_US/governments). Though there are other credit rating firms that operate in a number of countries, few can match the global reach of the “big three” in the triopoly of credit rating industry.

2. State governments and their instrumentalities in the United States, including cities, are rated on a separate U.S. Public Finance schedule, however.

3. Credit ratings serve other important roles as well, beyond the roles of information resolution and certification (e.g. Bongaerts et al., 2012; Johnson & Kriz, 2002). They are often said to exercise a monitoring role of city credit quality and compel discipline among politicians and policymakers (Bayoumi et al., 1995; Hackworth, 2007; Moldogaziev & Guzman, 2015), and are also used as policy benchmarks or regulatory mechanisms within the financial system (Majnoni et al., 2002; De Pascalis, 2017; Sinclair, 1994; White, 2010). Therefore, it is reasonable to conclude that a global rating firm, as an actor, and their ratings, as a product, have become an integral and indivisible part of urban governance and urban production around the world. In financialized urban environments, cities must seek to understand one of the very important financial intermediation actors–the credit rating firms.

4. The caveat here is that the provision of urban services with the help of the financial sector is not new, at least within the contexts of several countries (Hackworth, 2009; Hildreth & Zorn, 2005; Pagano & Perry, 2008). In the United States, cities successfully “worked with various financial actors to pay for brownfield redevelopment involving the attracting of creative, financial and media employment, residential gentrification of downtown, and the revitalization of post-industrial landscapes such as rail-yards and waterfronts” (Ward, 2017).

5. Additionally, researchers can estimate several models with different counts of topics and calculate their perplexity scores (Blei & Lafferty, 2007). Perplexity is a theoretical measure of how well a topic model with given parameters fits the corpus (Blei et al., 2003; Hollibaugh, 2019). The lower the perplexity score, the better fit to the data. Researchers may select the number of topics at which the marginal perplexity score stops decreasing (Blei & Lafferty, 2007). It should be noted, however, that parsimony is important here. Experiments show that topics become more fine-grained and less useful with larger numbers of estimated topics (Chang et al., 2009). Hence, the perplexity score should complement, rather than replace, substantive theoretical, and methodological meaning. In our case, perplexity scores show that models with topics greater than 10 are inferior models, but anything less than that is difficult to sort by model fit as the perplexity scores grow monotonically without any significant breaks in the distribution. Therefore, perplexity scores in this study are not particularly useful for selecting topic counts.