The Abstract
In the abstract, there is a restatement of a long-standing debate in sociology: is educational
expansion creating a devaluation around degrees (i.e., credential inflation theory) or is the
linkage between credentials and rewards determined by the skills they represent and degree tier?
Using a cross-country multilevel regression, the authors provide evidence that irrespective of
skills, lower-level degrees (e.g., associate or short-cycle) face devaluation when there are higher-
level equivalents classic credential inflation. In comparison, higher-level degrees (bachelor’s and
above) are only devalued when accompanied by lower levels of cognitive skills. Significantly,
high skills are not penalized for being without degrees in high-level expansion, which suggests
that at higher levels there is a shift toward skill-based valuation referred to by the authors as
decredentialization. The study provokes challenges to any conservative and singular
understandings of credential devaluation in educational stratifications. It provides a dual-process
account of educational stratification.
The Introduction
Araki and Kariya frame their work at the intersection of contradictory theoretical traditions:
- Credential inflation theory (Collins, 1979): education is a positional good whose yield
diminishes with excessive supply.
- Human capital theory (Becker, Mincer): education increases productivity and therefore merits
financial remuneration.
- Signaling theory (Spence): degrees signify trainability, not necessarily competence.
- Social closure (Weber, Bourdieu): elites rely on credentials to monopolize opportunity.
The authors identify two major gaps:
1. Conflation of credential levels: most research treats "tertiary education" as a single entity, and
many degrees (a bachelor's degree quite different from a short cycle diploma) occupy vastly
different positions in the labor market.
2. Not all degree holders share the same skill competency; some do not possess high enough
literacy or numeracy to back their degree, classifying them as "nominal" degrees.
Their key theoretical innovation is to suggest decredentialization a process where credentials lose
privilege and skills become more important when allocating rewards. This is not to revert to
functionalism but, rather a conditional meritocracy, to occur only under one specific structural
condition: massificaion of higher level tertiary education.
With tertiary achievement globally surging (38% of OECD adults holding tertiary qualifications
by 2019), policymakers and experts must ask themselves a thought provoking question: Is more
education better? And for whom? The response of the article is subtle: expansion has mixed
consequences depending on what kind of schooling is being expanded and who is doing the
expanding.
The Method
The authors utilize a thorough cross-national multilevel design using PIAAC data (N = 91,217
individuals across 26 countries). The methodological choices they made are innovative yet
defensible:
- Operationalization of skills: They employ PIAAC's standardized literacy and numeracy scores
(≥326 = "high skills"), a proxy for general cognitive ability established in prior work (Hanushek
& Woessmann). Non-cognitive skills are not included (a limitation they recognize), but PIAAC
remains the only dataset that allows for linking adult skills, credentials, and labor outcomes
internationally across multiple countries.
- Credential stratification: They are able to differentiate ISCED Level 5 (short-cycle tertiary)
from Level 6 + (bachelor's degree and above), allowing for examination of hierarchical
credential devaluation.
- Societal-level measures: They created two indicators of educational expansion (absolute
prevalence and inter-cohort growth) for each credential tier so that they can test whether scarcity
or rate of change produces devaluation.
- Analytical strategy: They progressively test their hypotheses by using four nested multilevel
models as well as including robustness checks (e.g., Model 4 does not include macro-controls).
A particularly strong move in this analysis is the use of cross-level interactions that shows how
macro-conditions are reshaping micro-rewards.
A few limitations:
Cross-sectional design: it cannot directly be used to draw conclusions regarding causation, and it
cannot be used for tracking individuals longitudinally.