From ministry to market: higher education expansion and state-designed marketization in Uzbekistan, 2015–2025

1 Introduction

The way a country pays for its universities is never only a question of budget administration. It defines where public responsibility ends and market exchange begins, and for that reason it is a classic problem of political economy. Following Caporaso and Levine (1992), we understand political economy as the study of how the state allocates resources and constructs markets, and not only of the consequences of such choices. Seen in this light, the recent higher education reforms in Uzbekistan are political-economy acts: the state has drawn a new boundary between the ministry and the market.

Uzbekistan is an instructive case for such a study. After independence in 1991 the country kept its centrally planned, ministry-financed university system, and the state governed the sector as one of the commanding heights of the economy (Ruziev and Burkhanov, 2018). The arrangement was typical of the developmental state, in which the government steers human capital formation in the interest of national development. Since 2017, under President Shavkat Mirziyoyev, a wave of reforms has been taking this system apart and rebuilding it. Our research question is therefore descriptive and institutional: how did the reforms of 2017–2025 reconfigure state–market relations in Uzbek higher education, and what do the official statistics tell us about the timing, the scale, and the character of this state-designed marketization?

Four bodies of literature structure the analysis, and from each of them we derive an observable proposition for the Uzbek case. First, the marketization literature has long argued that markets in higher education are politically constructed positional markets rather than spontaneous ones Marginson, (2013), and that neoliberal governance recasts universities as competitive enterprises of the knowledge economy (Olssen and Peters, 2005). The corresponding proposition (P1) is that the features of the market, such as entry, prices, and price ceilings, will be created and delimited by identifiable acts of the state. Second, the cost-sharing literature (Johnstone, 2004; Johnstone and Marcucci, 2010) predicts that fiscal pressure and enrollment expansion shift the financing burden from the government toward students and households; the proposition (P2) is that expansion will go together with a rising fee-paying share of enrollment. Third, the academic capitalism perspective (Slaughter and Leslie, 1997) expects institutions to pursue market and market-like revenues up to the limits that the state permits; the proposition (P3) is that institutions with pricing discretion will actually use it, up to the regulatory ceiling. Fourth, resource-dependence theory (Pfeffer and Salancik, 1978) predicts that as universities depend more and more on tuition, their behavior becomes aligned with the interests of that revenue source. Kholmuminov et al. (2019) documented such a dynamic for Uzbek public universities already in 2000–2013, and our proposition (P4) is that the reformed system institutionalizes tuition dependence both in law and in the financing mix. In the Discussion we return to each proposition and ask whether the Uzbek case confirms, modifies, or challenges the underlying theory, placing the case next to comparative evidence on quasi-markets and post-Soviet reform in the region (OECD, 2017; Ruziev and Burkhanov, 2018).

One distinction runs through the whole paper. The reforms consist of two parallel tracks that work through different mechanisms and at different times. The access track began in 2017–2018 and includes the relaxation of admission quotas, the licensing of new state, non-state, and foreign institutions, and the growth of distance learning. The finance track is anchored by Resolution PQ-61, adopted in December 2021 and in force from 1 January 2022, which delegated pricing authority and residual claims to designated universities. System-level enrollment dynamics speak mainly to the first track, whereas fee structures and the financing mix speak to the second. Our empirical strategy keeps the two apart. Following this logic, the enrollment series is analyzed in this paper as evidence on state-designed expansion, and not as an outcome measure of the self-financing resolution.

2 Methods

The legal-institutional analysis follows a documented procedure rather than a simple narrative chronology. The document corpus consists of the presidential decrees and resolutions and the constitutional provisions on higher education access, governance, and financing adopted between January 2017 and the retrieval date. Documents were identified through full-text search of the National Database of Legislation of the Republic of Uzbekistan (lex.uz) using Uzbek and Russian search terms (“oliy ta'lim,” “vysshee obrazovanie,” “moliyaviy mustaqillik,” “finansovaya samostoyatel”nost””) and were cross-checked against the annual reform reports of the Ministry of Higher Education, Science and Innovation. The inclusion criteria were: acts that establish or modify (i) entry and licensing of institutions, (ii) admission quotas, (iii) financing arrangements or financial autonomy, or (iv) governance of state universities. Acts concerning individual institutions' non-financial matters and acts on pre-tertiary education were excluded. Six acts met these criteria and are analyzed here: PP-4391 (11 July 2019), UP-5847 (8 October 2019), PQ-60 and PQ-61 (both of 24 December 2021, in force from 25 December 2021), the amendments to Annex 1 of PQ-61 (for example, PQ-290 of 21 June 2022), and Article 51 of the 2023 Constitution.

All documents were read in the Uzbek and Russian originals by the authors, who are fluent in both languages, and the quoted provisions were translated by the authors. One methodological point deserves emphasis. The online texts on lex.uz are consolidated texts that incorporate later amendments, so historical claims, such as the number of universities listed in Annex 1 of PQ-61 in a given year, were verified against the version of the act in force on the relevant date, using the date-versioning function of the database; retrieval dates are reported in the reference list. Each act was then coded on four dimensions of marketization that follow from the theoretical framework in Section 1: price authority, residual claims, entry and licensing, and governance. Where secondary reports and the legal text disagreed, we relied on the consolidated legal text, and our interpretations were triangulated against ministry announcements and the official statistics described in Section 2.2.

All quantitative data are official secondary data. Table 1 lists every variable together with its definition, period, source, and direct location. The enrollment and institution series come from the Statistics Agency of the Republic of Uzbekistan (stat.uz), release “Higher education in the Republic of Uzbekistan” of 28 May 2025, which reports headcounts at the start of each academic year. The series contain no missing values, and no data cleaning was needed beyond unit conversion; every value we use is reproduced in the note to Table 1.

Table 1. Variables, definitions, periods, and data sources. Direct document locations and retrieval dates are given in the reference list.

Enrollment series (students): 2015/16 = 264,300; 2016/17 = 268,300; 2017/18 = 297,700; 2018/19 = 360,200; 2019/20 = 441,000; 2020/21 = 571,500; 2021/22 = 808,400; 2022/23 = 1,042,100; 2023/24 = 1,314,500; 2024/25 = 1,432,800. HEIs (incl. branches): 69; 70; 72; 98; 119; 127; 154; 191; 219; 222, respectively. Non-state HEIs: 5; 17; 42; 93; 99 for 2020/21–2024/25. Each value is taken from the Statistics Agency release cited above.

Two official totals circulate for 2024/25, and they must not be confused with each other. The Statistics Agency reports 1,432,800 students at the start of the 2024/25 academic year (roughly September 2024). The statistical portal of the Ministry of Higher Education, Science and Innovation (stat.edu.uz) reports 1,551,172 students in a January 2025 snapshot (Ministry of Higher Education Science and Innovation of the Republic of Uzbekistan, 2025), of whom 173,788 held state grants and 1,377,384 studied under payment contracts; the same portal's breakdown by institutional ownership at that date sums to 1,511,185 (995,551 in state and 515,634 in non-state institutions), that is, about forty thousand less than the funding-basis total. The differences reflect different producers, different reference dates (start of year vs. mid-year, after additional contract admissions), and differences in coverage. For the trend analysis we therefore use only the Statistics Agency series, which is internally consistent across all 10 years, and we use the Ministry snapshot only for the funding-basis and ownership decompositions, always stating its January 2025 reference date. The fee-paying share of 88.8% is the ratio 1,377,384/1,551,172 within the Ministry snapshot, and it is never combined with the Statistics Agency total.

We also fix the terminology, since the concepts involved are related but not interchangeable. “Marketization” means the introduction of market features, such as price signals, competition, and private entry, into a state-governed sector. “Financial autonomy” means the legal authority of a state university under PQ-61 to set contract tuition, keep year-end balances, and manage its own funds. “Self-financing” means the funding regime in which a university's operations are financed from its own revenues, chiefly tuition, rather than from line-item ministerial allocations. Finally, the “fee-paying (contract) share” is a funding-status indicator: it measures how many students study under payment contracts, not who ultimately bears the cost, because contract tuition may be paid by households, employers, scholarships, or state-backed student loans. Each of these indicators is analyzed in its own right.

We use segmented log-linear trend models purely as descriptive devices for characterizing the timing and magnitude of the enrollment acceleration. We want to be very clear about this choice. With ten annual observations, several concurrent interventions, no comparison series, and a breakpoint that was itself searched over candidates, the analysis cannot support confirmatory hypothesis testing, and it cannot attribute anything causally to a particular decree. All p-values reported below are descriptive summaries of fit, not tests of reform effects, and the search over breakpoints means that the nominal significance levels overstate the evidence against a stable trend. The models describe the shape of the series; the institutional argument is carried by the legal analysis, not by the regression.

The dependent variable is the natural logarithm of total enrollment. The time variable t is coded 0 for 2015/16 and increases by one per academic year up to 9 for 2024/25. Model 1 is a simple log-linear trend, ln (Enrollment_t) = α + βt + ε_t. Model 2 is a segmented specification, ln (Enrollment_t) = α + β1t + β₂D_t + β3D_t(t – t0) + ε_t, where D_t = 1 for t ≥ t0 and 0 otherwise, β1 is the pre-break slope, β₂ the level change, and β3 the slope change. As the reference placement we take t0 = 4 (2019/20), the first academic year after the adoption of the Concept (UP-5847). Because this choice is a convention and not a finding, Model 2 is re-estimated with t0 ε {2, 3, 6}, corresponding to 2017/18, 2018/19, and 2021/22 (the year in which PQ-61 entered into force), and the complete coefficient sets are reported in Table 2 with 95% confidence intervals.

Table 2. Breakpoint sensitivity: complete coefficients of the segmented model for each candidate break, with 95% confidence intervals.

Source: authors' calculations from the series in Table 1. Cells report OLS point estimates with 95% confidence intervals in brackets. The pre-break segment for the 2017/18 placement contains only two observations, so its pre-break parameters are fitted essentially without residual degrees of freedom and are reported for completeness only. Because the breakpoint was searched over candidates, the nominal descriptive p-values overstate the evidence against a stable trend and are not interpreted as tests (Section 2.3). Break timing is treated as indeterminate within 2017/18–2019/20; the 2021/22 placement, corresponding to the entry into force of PQ-61, fits worst among the candidates.

Estimation is by ordinary least squares in Python 3.12.3 using statsmodels 0.14.6 (with NumPy 2.4.4 and SciPy 1.17.1); the replication script accompanies the article as Supplementary Material. For each model we report conventional OLS standard errors, Newey–West heteroskedasticity- and autocorrelation-consistent (HAC) standard errors with one lag, 95% confidence intervals, R², the Durbin–Watson statistic, the residual sum of squares, and AIC computed as n·ln(RSS/n) + 2k. We stress that with n = 10 neither conventional nor asymptotic HAC inference is reliable: positive residual autocorrelation can make conventional standard errors anti-conservative, and the HAC correction is an asymptotic one. Both are therefore reported for transparency and treated as descriptive. Residual diagnostics include the Durbin–Watson statistic (interpreted heuristically, since tabulated critical values do not exist for such a small sample), a Shapiro–Wilk test of residual normality, and inspection of residuals and fitted values, which the Supplementary script reproduces. A Chow-type F statistic comparing the segmented model with the single trend is reported as a descriptive measure of the improvement in fit, not as a test of a reform effect.

3 Results

The sequence of legal acts, all of which can be retrieved from the National Database of Legislation (lex.uz), is informative in itself. Presidential Resolution PP-4,391 of 11 July 2019 (President of the Republic of Uzbekistan, 2019a), on new management principles in higher and secondary specialized education, set the stage. Presidential Decree UP-5,847 of 8 October 2019 (President of the Republic of Uzbekistan, 2019b) then approved the Concept for the Development of the Higher Education System until 2030, with the targets of raising coverage above 50% and placing at least ten universities among the top 1,000 of internationally recognized rankings. The decisive financial step came 10 years later. Resolution PQ-61 of 24 December 2021, in force from 25 December 2021, granted financial autonomy to the state universities designated in its Annex 1, effective from 1 January 2022, and the companion (President of the Republic of Uzbekistan, 2021b). Resolution PQ-60 of the same date extended academic and organizational-managerial autonomy to them (President of the Republic of Uzbekistan, 2021a).

The operative provisions of PQ-61 can be cited precisely. Clause 2 authorizes the listed universities to set contract tuition independently “with regard to demand in the market of educational services,” to hire domestic and foreign faculty on contractual terms, to award their own scholarships from their own funds, and to set performance indicators and remuneration. Clause 3 moves state financing to a per-student normative basis, transfers state-order funds to the university's extra-budgetary treasury sub-account, and stipulates that the year-end balance remains at the disposal of the university; this residual-claim provision is, in our reading, the heart of the self-financing model. Clause 5 caps contract tuition at no more than three times the base contract cost for the corresponding field of study and permits differentiated pricing. Clause 6 assigns to supervisory boards the approval of strategic development plans, business plans, and contract-admission parameters. Annex 1 designated 35 institutions in the version in force during 2022. The note to Annex 1, as amended, delegates further additions from 1 January 2023 to the Ministry of Higher Education, Science and Innovation in agreement with the Republican Council for Higher Education and the Ministry of Economy and Finance, and successive amendments (for example, Resolution PQ-290 of 21 June 2022) have expanded the list to 41 institutions in the version of Annex 1 in force at the time of retrieval (2026). Finally, after the constitutional referendum of 30 April 2023, Article 51 of the Constitution (Constitution of the Republic of Uzbekistan, 2023) guarantees competitive state-funded access to higher education and enshrines the academic freedom, self-governance, and freedom of research and teaching of higher educational institutions.

If we read these acts against the coding framework of Section 2.1, every parameter of the new market turns out to be located in the hands of the state: entry (licensing of state, non-state, and foreign providers), the base price (set administratively), the price ceiling (three times the base), residual claims (delegated to universities), and membership of the autonomous “club” (a state-controlled list). This is proposition P1 in legal form: the features of the market were created and delimited by identifiable acts of the state.

Total tertiary enrollment rose from 264,300 in 2015/16 to 1,432,800 in 2024/25, a 5.4-fold increase, and the number of higher education institutions including branches grew from 69 to 222 over the same years (Table 1; Figure 1). Two contextual series help to clarify what kind of expansion this was, since they describe the demand-side conditions for it. The first concerns demography. The expansion is participation-driven rather than demographic: the World Bank gross tertiary enrollment ratio rose from 8.4% in 2016 to 45.8% in 2023, and the Statistics Agency reports coverage of the 18–23 age group of 47.7% for 2024/25, while the underlying youth cohort changed only modestly over the decade in this young population (the median age was 27.9 years in 2024; United Nations Department of Economic and Social Affairs Population Division, 2024). A five-fold increase in enrollment against a roughly stable cohort implies that nearly all of the growth reflects rising participation made possible by policy. The second concerns household capacity to pay, which grew much more slowly than enrollment. The average monthly nominal wage rose from about 1.45 million soums in 2017 to about 5.36 million soums in 2024 (State Statistics Agency of the Republic of Uzbekistan, 2025), a 3.7-fold nominal increase over a period in which enrollment grew 4.8-fold and the fee-paying share moved toward nine-tenths. Against an average annual wage of roughly 64 million soums in 2024, the base contract fee for 2024/25 corresponded to about 12% (Pedagogy, 7.4 million soums) to 18% (Law, 11.25 million soums) of the average annual wage, whereas tuition at the ceiling, 29.4 million soums at the University of World Economy and Diplomacy, corresponded to roughly 46% of it (Table 3).

Table 3. Marketization, financing-mix, and demand-side indicators, with exact values, reference dates, and sources.

The fee-paying share is a funding-status indicator (who holds a payment contract), not a measure of who ultimately bears the cost: contract tuition may be paid by households, employers, scholarships, or state-backed loans. Ministry-portal figures refer to the January 2025 snapshot and are not comparable with the Statistics Agency start-of-year series used in Tables 1, 2, 4 (Section 2.2). Values previously reported as approximate have been replaced with documented figures or removed; the previously reported decline in the state-funded share of admissions (32.8% in 2016 to 15.7% in 2023) has been removed because a directly citable official series could not be identified.

The ownership structure of supply shifted as well. The number of non-governmental HEIs grew from 5 in 2020/21 to 99 in 2024/25 (the annual series is 5, 17, 42, 93, 99; Statistics Agency), and there were also 30 foreign universities and branches. According to the Ministry's January 2025 snapshot, 995,551 students (65.9%) studied in state institutions and 515,634 (34.1%) in non-state institutions (Ministry of Higher Education Science and Innovation of the Republic of Uzbekistan, 2025). Marketization in Uzbekistan thus goes forward along two lines at once, through private entry and through the internal transformation of the state sector, a pattern that is familiar from other post-Soviet systems (OECD, 2017; Ruziev and Burkhanov, 2018).

Table 4 reports the two reference models, and Table 2 the full breakpoint sensitivity results. For Model 1 the trend coefficient is 0.2139 (SE = 0.0132; HAC SE = 0.0140; 95% CI 0.1834–0.2443), which implies average annual growth of about 23.8%, with R² = 0.970. The Durbin–Watson statistic of 0.79, read heuristically, points to strong positive residual autocorrelation: a single straight line systematically misses the acceleration in the middle of the period, and this is what motivates the segmented description.

Table 4. Descriptive trend models. dependent variable: ln (total enrollment), 2015/16–2024/25 (n = 10).

Source: authors' calculations from the series in Table 1. Model 1: R² = 0.970, adjusted R² = 0.967, Durbin–Watson = 0.79, RSS = 0.1149, AIC = −40.7. Model 2: R² = 0.991, adjusted R² = 0.987, Durbin–Watson = 1.74, RSS = 0.0345, AIC = −48.7; Shapiro–Wilk on residuals W = 0.867, p = 0.091; Chow-type F _(2.6) = 6.98, descriptive p = 0.027. HAC = Newey–West standard errors (lag 1), asymptotic and reported for transparency only. AIC = n·ln (RSS/n) + 2k. β = 0.2139 corresponds to average annual growth of exp (0.2139) – 1 ≈ 23.8%; the post-break slope of 0.2470 corresponds to ≈ 28.0% vs. ≈ 10.9% pre-break. All p-values are descriptive summaries of fit, not tests of reform effects (Section 2.3).

For Model 2 with the reference break at 2019/20, the pre-break slope is 0.1033 (95% CI 0.0203–0.1863), or about 10.9% annual growth over 2015/16–2018/19. The slope change is 0.1437 (95% CI 0.0496–0.2378), which gives a post-break slope of 0.2470 (95% CI 0.2026–0.2914), or about 28.0% annual growth. The level shift of 0.191 is estimated imprecisely (95% CI −0.073 to 0.455). R² rises to 0.991, the Durbin–Watson statistic to 1.74, and a Shapiro–Wilk test does not reject residual normality (W = 0.867, p = 0.091). The Chow-type comparison with the single trend gives F _(2.6) = 6.98, with a descriptive p of 0.027.

The sensitivity analysis in Table 2 supplies the essential caveat, and, for the question as we have now framed it, also the essential finding. Placing the break at 2017/18 (AIC −50.6) or at 2018/19 (AIC −50.7) fits somewhat better than 2019/20 (AIC −48.7), while placing it at 2021/22, the year the financial-autonomy resolution entered into force, fits worst among the candidates (AIC −47.2), with a slope-change estimate that is small and imprecise (0.0367; 95% CI −0.0645 to 0.1379). In other words, the data describe an acceleration that began with the access-expansion measures of 2017–2019 and that cannot be narrowed to a single year inside that window; they give no support to dating the acceleration to the 2022 self-financing reform. Year-on-year growth peaked at 41.5% in 2021/22 and fell to 9.0% by 2024/25 as the base grew larger. Consistent with Section 2.3, no conclusion in this paper hinges on any reported p-value.

The finance track becomes visible not in the enrollment series but in prices and in the funding mix (Table 3). Base contract fees, which had been left unchanged for 4 years, were raised by about 15% on average for 2024/25 (Pedagogy from 6,412,610 to 7,400,000 soums; Law from 9,809,260 to 11,250,000 soums; Ministry of Higher Education Science and Innovation of the Republic of Uzbekistan, 2024). Universities with financial autonomy went further within the limits of the PQ-61 cap. At the University of World Economy and Diplomacy, one of the listed autonomous universities, the supervisory board set first-year undergraduate tuition for 2024/25 at 29,428,000 soums, which is exactly three times the base of 9.8 million soums for the field, in other words at the regulatory ceiling (University of World Economy and Diplomacy, 2024). The state defined the maximum, and the university priced at that maximum. This is the observable implication of proposition P3, although we would stress that evidence from a single elite institution illustrates the mechanism rather than establishes system-wide behavior. As for the funding mix, 88.8% of students in the Ministry's January 2025 snapshot studied on a fee-paying (contract) basis and 11.2% held state grants (Ministry of Higher Education Science and Innovation of the Republic of Uzbekistan, 2025). We repeat that this is a funding-status indicator: it does not show that students “finance their own education,” since contracts may be paid by households, employers, scholarships, or loans. Longitudinal data on the composition of university revenues and on government appropriations specifically for higher education have not yet been published. The available fragments, namely higher education's share of about 12% of the consolidated education budget in 2017 (World Bank, 2022), total public education expenditure of 5.47% of GDP in 2023, which is a whole-of-education figure that includes but does not isolate higher education (World Bank, 2025a), and the launch of a state-backed student-financing program in December 2025 (World Bank, 2025b), are reported in Table 3 as context and not as measures of the higher education financing mix.

4 Discussion

What the evidence describes is a restructuring of the developmental state rather than its retreat. The state did not withdraw from the sector; it changed its instruments. By keeping licensing, the base fee, the fee ceiling, and control over membership of the autonomous list in its own hands, the government repositioned itself from direct financier to designer and regulator of a quasi-market. In such a setting, the instruments of competitiveness, such as ranking targets, foreign branches, and supervisory-board pricing, are tools of governance that mediate between state and market, and not neutral management techniques.

Returning to the four propositions of Section 1 helps to discipline the theoretical contribution. P1, on politically constructed markets, is confirmed in an unusually legible form: every parameter of the market can be traced to a dated legal act. This supports the argument of Marginson (2013) that higher education markets are artifacts of the state, and it extends the argument to an authoritarian developmental context in which the construction is explicit rather than emergent. P2, on cost-sharing, is confirmed in its direction, since the expansion went together with a fee-paying share approaching nine-tenths; at the same time, the Uzbek sequence modifies the standard account (Johnstone, 2004), because here cost-sharing was not a reluctant fiscal adaptation but a designed feature of the expansion itself. P3, on academic capitalism, is illustrated rather than established: the one autonomous university for which official pricing documentation is available priced exactly at the state ceiling, which is consistent with revenue-seeking up to the permitted boundary (Slaughter and Leslie, 1997), but institution-level data across the autonomous list are needed before any system-wide claim can be made. P4, on resource dependence, is institutionalized in law: the residual-claim and pricing provisions of PQ-61 legally tie university finances to tuition markets, which is consistent with, and in a sense a legal deepening of, the tuition dependence that Kholmuminov et al. (2019) measured for 2000–2013; whether behavioral alignment followed remains an open empirical question. In the terminology of Caporaso and Levine (1992), the case is an authoritative allocation that manufactures a market out of a ministry, and it therefore challenges any reading of marketization as simple state withdrawal.

The descriptive trend results should be read within the two-track distinction. The sensitivity analysis locates the acceleration of enrollment growth in a window, 2017/18 to 2019/20, that saw quota liberalization, the licensing of new providers, the expansion of part-time and distance learning, and the adoption of the Concept, with the COVID-19 pandemic overlapping later; the financial-autonomy resolution of 2021 comes after the acceleration and is associated with no additional break in the series. For this reason we attribute the expansion to the access track of the policy regime as a whole and draw no conclusions about the self-financing reform from the enrollment data. Our claims about the finance track rest only on the legal provisions and on the pricing and funding-mix evidence of Section 3.4.

The reform is relevant to SDG 8, through the deepening of human capital, and to SDG 16, through the constitutional formalization of institutional autonomy (Article 51), and it connects to the emerging work on governance and sustainability models in Uzbek universities (Juraev et al., 2026). The same data, however, raise an equity concern, which we state as a hypothesis and not as a finding, because we do not observe household-level affordability, socioeconomic participation, debt, or dropout. Where 88.8% of students study on contracts, where tuition at the top of the permitted range rises faster than wages, and where grant coverage is thin, access plausibly becomes more and more conditional on the ability to pay, in line with the positional-goods logic of marketization (Marginson, 2013). The government's own launch of the student-financing program in December 2025 suggests that policymakers share this concern (World Bank, 2025b). Testing the hypothesis will require distributional data that do not yet exist publicly.

5 Limitations

The limitations deserve to be repeated rather than hidden. Ten annual observations can describe a growth pattern, but they cannot identify the effect of any particular act. The breakpoint is statistically indeterminate within a 3-year window, and searching over breakpoints inflates the apparent significance. The level-shift parameter is imprecise, and neither conventional nor HAC inference is reliable at this sample size. The enrollment surge bundles together quota relaxation, supply expansion, distance learning, pandemic effects, demography, and the financing reform. The fee-paying share is a funding-status measure observed at one date and on a different accounting basis from the trend series. The documentary analysis, though version-controlled, covers presidential-level acts and does not extend to ministerial implementing regulations. The decisive next step, in our view, is institution-level panel data comparing autonomous and non-autonomous universities around 1 January 2022, covering tuition levels, revenue composition, and admission behavior, joined to an affordability analysis of contract fees relative to household incomes; such data are not yet consistently and openly available for the post-2019 years.

6 Conclusion

The Uzbek state did not find a market in higher education; it built one of its own design. Enrollment expanded 5.4-fold over a decade, 88.8% of students now study on a fee-paying basis, and the growth trajectory shifted upward during the reform period, a regime change that our descriptive models date to the access reforms of 2017–2019 while declining to credit any single decree. The clearest evidence of the new financial regime lies in prices: the state set a ceiling of three times the base fee, and the flagship autonomous university priced exactly at it. Every parameter of this market, entry, the base price, the permissible mark-up, and membership of the autonomous list, remains in the hands of the state. That is why we read the reform as a transformation of the developmental state and not as its retreat.

Two implications follow. For researchers, competitiveness and marketing in systems of this kind are instruments of governance, and studying them outside their political-economy context misses most of the story; the priority is institution-level panel evidence on whether autonomy changed university behavior or only its funding channel. For policymakers, a system in which access increasingly runs through fee-charging institutions will sooner or later meet the limits of household capacity to pay, a constraint the government has implicitly acknowledged with its new student-financing scheme. Whether the quasi-market broadens opportunity or merely re-prices it will depend on choices that are still to be made about grants, fee ceilings, and accountability.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Mihaela Tofan, Alexandru Ioan Cuza University of Iasi, Romania

Reviewed by: Sisca Septiani, Universitas Ngudi Waluyo, Indonesia

Olha Hrynkevych, Ivan Franko National University of Lviv, Ukraine