Introduction
In the wake of the Covid-19 pandemic, Indonesia’s education policy discourse has been dominated by a singular, uniform narrative: recovery. Macro evaluations consistently parade administrative metrics as definitive proof that curriculum overhauls and digital interventions have successfully tamed the ghost of learning loss.
Yet, has this cognitive recovery truly trickled down equitably to the grassroots? Or is it merely a statistical illusion—a macro-average that masks the permanent scarring inflicted upon vulnerable student cohorts?
This article seeks to dissect the underlying mechanisms operating within the “black box” of national education policy. By analyzing longitudinal panel data comprising 87,696 observations of public primary schools (SD/MI) across Indonesia from 2021 to 2025, we uncover a deeply concerning structural anomaly. Rather than fostering equity, current recovery frameworks are actively reactivating socioeconomic stratification as the primary engine driving cognitive inequality among the nation’s youth.
To illustrate the structural nature of this data empirically, the table below provides a summary of the panel regression estimates from the two core models undergirding this critique.
Table 1. Fixed Effects (FE) Panel Regression Estimates
| (variables) | LIT (Literacy) | NUM (Numeracy) |
|---|---|---|
| 2021.year (Baseline) | 0.000 | 0.000 |
| (.) | (.) | |
| 2023.year | 6.148*** | 7.151*** |
| (0.457) | (0.397) | |
| 2024.year | 7.043*** | 13.796*** |
| (0.486) | (0.423) | |
| 2025.year | 4.380*** | 8.858*** |
| (0.478) | (0.416) | |
| SES_school | -0.040*** | -0.049*** |
| (0.009) | (0.008) | |
| 2021.year#c.SES_school | 0.000 | 0.000 |
| (.) | (.) | |
| 2023.year#c.SES_school | 0.048*** | 0.056*** |
| (0.007) | (0.006) | |
| 2024.year#c.SES_school | 0.066*** | 0.065*** |
| (0.007) | (0.006) | |
| 2025.year#c.SES_school | 0.054*** | 0.074*** |
| (0.007) | (0.006) | |
| 1.status_wilayah_num (Baseline) | 0.000 | 0.000 |
| (.) | (.) | |
| 2.status_wilayah_num (RURAL) | 1.241 | 1.269 |
| (2.183) | (1.898) | |
| 3.status_wilayah_num (URBAN) | -7.694 | -0.407 |
| (7.507) | (6.529) | |
| 1.jenis_sek_num | 0.000 | 0.000 |
| (.) | (.) | |
| 2.jenis_sek_num (SD) | -4.956 | -6.982 |
| (9.873) | (8.587) | |
| _cons | 54.191*** | 42.560*** |
| (10.099) | (8.784) | |
| N (Observations) | 87,701 | 87,696 |
| R-squared (Within) | 0.257 | 0.520 |
The empirical estimates above confirm the existence of a massive asymmetric recovery. The interaction terms between chronological time and institutional socioeconomic capacity (year#c.SES_school) are positive and highly statistically significant (p < 0.01). This proves that the velocity of post-pandemic recovery is fundamentally dictated by a school’s economic capital. Furthermore, the R2 within for the Numeracy Model reaches 52.0% (0.520), validating the robustness of this FE panel model in capturing the internal dynamics of cognitive changes within educational institutions over time.
Crucially, a compelling statistical divergence emerges when comparing the R2 within coefficients: the Literacy model yields a far lower explanatory power (25.7%) than its Numeracy counterpart (52.0%). From a sociological perspective, this structural mismatch reveals that literacy is fundamentally a culture-dependent competency, intensely dictated by unobserved external variables such as home literacy environments and parental cultural capital. Conversely, numeracy functions as a strictly school-dependent skill, where learning gains are directly anchored to formal classroom instruction. The muted explanatory strength of the literacy model underscores a critical systemic truth: the state’s administrative recovery framework remains profoundly decoupled from the primary upstream causal drivers of children’s reading competencies.
1. The Asymmetric Recovery: The Mechanics of Exclusion
The robust econometric findings extracted from this five-year panel dataset dismantle the state’s narrative of universal recovery. Through Fixed Effects estimation—which rigorously controls for all time-invariant, unobserved school characteristics—the interaction variables between time and a school’s Socioeconomic Status (year#SES_school) yield positive coefficients that are highly statistically significant (p < 0.001) across both literacy and numeracy outcomes.
From an ontological perspective, this reveals a deeply entrenched asymmetric causal mechanism. While the pandemic acted as a universal external shock, the structural capacity to bounce back was fiercely dictated by socioeconomic capital. Affluent schools operated within a resilient ecosystem: parents could seamlessly substitute formal schooling with private resources, digital access remained uninterrupted, and teachers adapted swiftly. Consequently, once lockdowns vanished, these schools accelerated.
Conversely, for children trapped in under-resourced schools, the disruption of formal classroom instruction served as a definitive sentence of long-term cognitive regression. Generic, “one-size-fits-all” recovery interventions failed to engage with these underlying causal structures. As a result, rather than narrowing, the cognitive divide across social strata has widened brutally in the post-pandemic era.
2. Unmasking the Rural Crisis: Spatial Disadvantage vs. Structural Deprivation
Some observers might misinterpret the empirical results in Table 1, noting that the physical-geographical variable for status_wilayah (Rural vs. Urban) appears statistically insignificant on its own (p > 0.05). Does this imply that rural districts did not suffer? Quite the contrary. This finding delivers the sharpest econometric indictment against conventional bureaucratic diagnostics.
Fixed Effects models operate by isolating the pure, independent impact of a variable. Once the profound destructive power of institutional poverty (SES_school) is accounted for simultaneously, the independent RURAL dummy loses its statistical significance. Why? Because the entire spectrum of suffering, isolation, and systemic underperformance in Indonesia’s rural districts has already been fully absorbed by structural economic deprivation (manifested as an exceptionally low SES_school baseline).
Rural schools are lagging post-pandemic not because of their physical coordinates on a map, but because rural spaces serve as geographic clusters for schools with the absolute lowest economic capital in the country. Attempting to resolve this crisis by merely distributing physical aid horizontally across spaces (Rural vs. Urban) without executing asymmetric, wealth-targeted resource redistribution is an academic and policy failure. Rural education is paralyzed because its underlying economic framework has been structurally suffocated by the system.
3. The 2024 Stimulus Spike and the 2025 Stagnation
The third anomaly documented in Table 1 reveals a jarring trajectory in the primary chronological coefficients. Using 2021/2022 as the nadir baseline, national cognitive scores—most notably in numeracy—experienced an aggressive surge in 2024, characterized by an exceptionally high coefficient. However, by 2025, this momentum collapsed significantly. A similar inverted U-curve pattern materialized within student literacy outcomes. Why did this cognitive recovery trajectory suffer such a volatile downturn by 2025?
This phenomenon exposes what can be theorized as a policy-driven “Sugar Rush Effect.” The massive influx of capital earmarked for school digitalization, hardware procurement, internet subsidies, and the administrative enforcement of a novel curriculum in 2023–2024 manufactured an artificial, cosmetic bump in short-term standardized assessments.
Yet, because these superficial interventions left the deep-rooted causal engines of cognitive growth untouched—such as genuine pedagogical capacity building for educators and nutritional security for impoverished children—the policy’s momentum evaporated by 2025. The state succeeded in keeping schools micro-occupied with administrative compliance checklists, but utterly failed to cultivate sustainable learning resilience within the students.
Policy Recommendations
If the panel data trajectory captured between 2021 and 2025 continues to be overlooked to protect bureaucratic egos and superficial programmatic victories, Indonesia is marching toward a silent intellectual catastrophe. We are actively manufacturing a generation that is administratively certified as graduated under a modern curriculum, yet fundamentally lacks foundational numeracy and critical reading competencies—with rural districts acting as the primary causal casualties.
Grounded in this Critical Realist analysis, two radical policy structural shifts must be executed immediately:
- Dismantle Geography-Centric “One-Size-Fits-All” Allocation: Resource allocation frameworks must no longer rely on superficial urban-rural divides that distort reality. The state must deploy a heavily asymmetric affirmative policy—funneling concentrated budgets explicitly into transforming the economic capacity of schools occupying the lowest socioeconomic deciles, which are disproportionately concentrated in rural districts.
- Institute a Moratorium on Cosmetic Hardware Projects: Pivot trillions of rupiah currently wasted on performative digital gadgets and bureaucratic software applications directly back toward upstream causal inputs. This means funding long-term, face-to-face intensive pedagogical training for teachers focusing on foundational literacy and numeracy, alongside robust nutritional support systems for impoverished student bodies.
It is time for policymakers to awaken from the complacency of macro-statistical illusions. The data has spoken with uncompromising clarity, and the choice before us is stark: restructure these underlying causal systems today, or harvest the structural failure of an entire generation tomorrow.