CA-TPU Quarterly Report (Q2 2026)

Key Points

In the second quarter of 2026 the CA-TPU index averaged 161.16 for the quarter, down 3.1% from the previous quarter (Q1 2026: 166.34) and sharply higher, by 40.7%, than a year earlier (Q2 2025: 114.56), ranking 4th among the 70 complete quarters since 2009. The more emblematic figure this quarter, however, is the quarterly average MA3 — 163.93, an all-time high, surpassing the 155.80 of Q2 2020. This is also the only quarter since the index began in which all three months sat in the MA3 ≥ 150 high-pressure zone (Q2 2020 had only two).

Within the quarter the path was a single-peak inverted V: 148.68 in April, a surge to 187.46 in May (the high of the quarter, and the peak of the current tariff war to date), then a retreat to 147.33 in June. The quarter collected 31,422 news articles, with 833 AND hits, essentially flat against the previous quarter (31,228 articles / 831 hits) — the scale of the data has moved from the explosive expansion of the first quarter onto a plateau. In terms of what drives the index, Uzbekistan has displaced Kazakhstan as the country with the highest concentration of uncertainty, while dispersion across the four countries continues to converge.

1. Index Overview

The CA-TPU index averaged 161.16 over the quarter, down 5.18 points (−3.1%) quarter on quarter and up 46.60 points (+40.7%) year on year. The quarterly average MA3 was 163.93, up 18.43 points (+12.7%) quarter on quarter and up 52.78 points (+47.5%) year on year. Note that the quarterly average index edged down while the quarterly average MA3 rose clearly — a divergence which shows that although the quarter set no new record for single-quarter intensity, it carried over in full the high pressure accumulated at the end of the previous quarter.

Table 1 CA-TPU Quarterly Index Comparison

Source: CA-TPU Index Project, aggregated from the monthly data published at catpu.caiees.cn.

Table 2 Monthly Path Within the Quarter

Source: CA-TPU Index Project.

The shape within the quarter is a clear single-peak inverted V: April fell back from the March high (183.48), May rebounded strongly to 187.46 and set the highest single-month reading of the current tariff war, and June retreated again. This differs from the “twin-peak” structure of the first quarter (175.43 in January, 183.48 in March), indicating that the discourse pressure of the second quarter was concentrated in one shock rather than in repeated oscillation. The MA3 readings for the three months were 157.43, 173.21 and 161.16 — all above 150, which is the single most important structural fact of the quarter.

2. Weekly Trajectory and the Rhythm of the Shock

To observe the micro-rhythm of the shock within the quarter, the project team aggregated the four countries’ combined AND hits and f_ratio (AND hits as a share of OR hits) by calendar week. The table below covers all 14 calendar weeks that intersect the quarter; the first and last of them straddle the quarter boundary and are flagged accordingly.

Table 3 Weekly Trajectory Within the Quarter (four countries combined)

Source: CA-TPU Index Project; f_ratio = AND hits / OR hits.

The weekly data reveal one clear downward thread: from the quarter’s peak of 6.23% in early April, the f_ratio slid almost monotonically to 3.81% in late June, a fall of about 39% within the quarter. Only one pronounced rebound interrupted it — the week of 8 June recovered to 5.28% to form a secondary high — but it did not change the overall trend.

One phenomenon here calls for explicit comment, because it invites misreading: the combined f_ratio peaked in early April whereas the CA-TPU index peaked in May, and the two are not synchronised. The reason lies in how the two are aggregated. The combined f_ratio is a simple unweighted pooling of the four countries and is therefore dominated by Kazakhstan and Kyrgyzstan, which generate the largest news volumes; the CA-TPU index, by contrast, first standardises each country separately and then composites the sub-indices by trade weight, so Uzbekistan’s component — which surged in May — pushed the composite to its peak in that month. The two indicators measure different facets of the same phenomenon and should not be used to validate each other.

3. In-Depth Analysis of the News Data

3.1 Quarterly totals

Table 4 Quarterly News-Data Totals

Source: CA-TPU Index Project.

The defining feature of the data scale this quarter is “double year on year, flat quarter on quarter.” Against the same quarter last year, total news volume grew 108.7% and AND hits grew 72.5%, extending the twin effects of an expanded collection system and a genuine strengthening of the signal. Against the previous quarter, however, all three scale indicators moved by less than 2%, showing that the dividend from expansion has been fully released and the sample size has entered a stable plateau. This matters for interpretation: the jump in scale reported in the first-quarter edition as the “highest single quarter on record” was a one-off event, and the high year-on-year growth rates will now gradually fall back toward the growth rate of the genuine signal.

The combined f_ratio edged down from 5.16% in the previous quarter to 5.09%, essentially unchanged, but remains well above the 4.38% of the same quarter last year. With the sample no longer expanding, a stable f_ratio means that the output density of “high-value news” has settled at a high level.

3.2 Cross-country comparison

Table 5 Four-Country Quarterly Comparison (year on year)

Source: CA-TPU Index Project.

The sharpest change is Uzbekistan. Its AND hits rose from 70 in the same quarter last year to 215, up 207.1% year on year, while its f_ratio climbed from 0.0375 to 0.0985, an increase of more than 1.6 times, making it the country with the highest concentration of uncertainty of the four. This is not an artefact of sample expansion: the year-on-year growth in the country’s total quarterly news volume (about 38%) is far below the growth in AND hits, which points to a genuine rise in signal density.

Kazakhstan shows the classic pattern of “volume up, ratio down.” Its f_ratio fell steeply from 0.1192 to 0.0488, a decline of 59%, which on the surface looks like a marked drop in attention; yet over the same period the absolute number of its AND hits rose from 247 to 324, an increase of 31.2%. The real cause lies in the denominator: the country’s total quarterly news volume jumped from 2,519 articles to 11,893, a rise of 372%, so news volume expanded far faster than high-value news and diluted the ratio. Kazakhstan must be read through absolute counts and ratios together; reading the f_ratio alone would yield a conclusion opposite to the facts.

The changes in Kyrgyzstan and Tajikistan are milder but point the same way. Kyrgyzstan’s AND hits rose 66.4% year on year and its f_ratio moved from 0.0204 to 0.0312, still the lowest absolute level of the four; Tajikistan’s AND hits rose 98.2% and its f_ratio reached 0.0664, second among the four, but its sample base is the smallest of the four and its readings the most volatile, so they should be read with caution.

3.3 Month-by-month evolution within the quarter

Table 6 Month-by-Month Evolution of the Four Countries’ f_ratio

Source: CA-TPU Index Project.

The evolution within the quarter further confirms the source of the May peak: of the four countries, only Uzbekistan saw a significant lift in its f_ratio in May (0.0964 → 0.1169); the other three either flattened, declined or bottomed out in that month. This explains why the composite index surged in May while the combined f_ratio kept falling in the same month — the peak was driven by a single country, not by a synchronised regional move.

4. Media Source Analysis

The eight main outlets across the four countries account for 833 AND hits in total this quarter. The table below ranks them by contribution and reports each outlet’s f_ratio, so as to distinguish “high output” sources from “high concentration” ones.

Table 7 TPU Contribution of the Eight Outlets

Source: CA-TPU Index Project; share = the outlet’s AND hits divided by the four-country total of AND hits.

On concentration, forbes.kz (0.1255) and uza.uz (0.1174) lead clearly, and the two values are strikingly close — a fact worth pausing on. The former is a market-oriented specialist business outlet, the latter a state news agency; their editorial logics are entirely different, yet they converge on the density with which trade, policy and uncertainty appear together. This suggests that policy uncertainty is now salient enough that neither the professional financial perspective nor the official-announcement perspective can avoid it, so the overlap between their agendas has increased.

On output, inform.kz ranks first with 200 articles (24.0%), yet its f_ratio is only 0.0353 — the signature of a general portal: the largest total news volume (10,548 articles) but a relatively low density of high-value content. Kyrgyzstan’s 24.kg is more extreme still, with an f_ratio of just 0.0248, the lowest of the eight. The top three outlets (inform.kz, uza.uz, forbes.kz) together contribute 61.1% of AND hits, a fairly high degree of concentration; this means the index carries a certain sensitivity to the editorial orientation of these three, and whether their article mix changes systematically will require continued monitoring.

5. Keyword Analysis

5.1 High-frequency keywords this quarter

After merging Russian case forms, the keyword counts across all high-value news of the quarter give the following high-frequency lemma families in the three categories.

Category T (trade), top five: торговля (trade, 1,208); соглашение (agreement, 1,053); экспорт (exports, 628); санкции (sanctions, 517); договор (treaty, 315).

Category P (policy), top five: решение (decisions, 988); политика (policy, 601); правительство (government, 492); министерство (ministries, 455); закон (law, 435).

Category U (uncertainty), top five: риск (risk, 1,150); кризис (crisis, 342); напряжённость (tension, 229); нестабильность (instability, 140); неопределённость (uncertainty, 98).

5.2 Differences in national agendas

Table 8 Core Agendas of the Four Countries

Source: CA-TPU Index Project; frequencies from four-country high-value news, after lemmatisation.

Kazakhstan’s agenda rests on the twin cores of “exports” and “risk.” Its risk family reaches 498 mentions, the highest of the four countries and 43.3% of the regional total for that family; exports, at 226, also leads the four. As the region’s largest economy and a major resource exporter, Kazakhstan’s sensitivity to changes in trading conditions translates directly into a high density of risk language; yet its core policy words remain the routine “decisions” and “policy,” with no clear signal of institutional adjustment as yet.

Uzbekistan displays a pronounced “agreement-led” profile. Its agreement family reaches 715 mentions, or 67.9% of the regional total for that family (1,053), far above its share of news volume, and it is the only one of the four countries whose leading core word is “agreement.” This corroborates its rise to first place on f_ratio: the country is in a period of intensive framework negotiation and institutional alignment, and its uncertainty arises from rules currently being renegotiated — an actively managed form of uncertainty.

Kyrgyzstan and Tajikistan represent two very different small-country paths. Kyrgyzstan is the only one of the four whose leading category-T word is “sanctions” (196 mentions), reflecting its structural position as a conduit for Russian re-export trade and its high sensitivity to secondary sanctions — a textbook case of sanctions spillover. Tajikistan’s second core policy word is “strategy” (55 mentions), unique among the four; its agenda leans toward medium- and long-term planning and structural arrangements rather than immediate bargaining. Given the limited sample size, however, the robustness of this feature awaits confirmation in later quarters.

5.3 Year-on-year change in keywords

Table 9 Year-on-Year Change in Major Lemma Families (Q2 2026 vs Q2 2025)

Source: CA-TPU Index Project; frequencies from four-country high-value news, after lemmatisation.

Looking at category totals, category P (policy) rose from 1,788 mentions in the same quarter last year to 4,218, up 135.9% year on year, far outpacing category T (3,112 → 4,320, +38.8%) and category U (1,147 → 2,017, +75.8%). This structural change extends and reinforces the central judgement of the first-quarter edition: policy response is itself becoming a new source of uncertainty. When government, ministries, law and strategy all grow faster than trade vocabulary, the market’s centre of attention has moved from “what will trading conditions be” to “how will each government respond.”

The item in this table that most needs explaining is the steep year-on-year fall in the two families for tariffs and duties (−66.5% and −80.7% respectively). That tariff vocabulary should decline during a period commonly called a tariff war looks contradictory, but is not. Q2 2025 was precisely the phase in which tariff measures were being introduced in quick succession and the media focused on the rates themselves; by Q2 2026 the centre of the debate had moved up a level — from “what is the rate” to “how are agreements to be negotiated, how are sanctions to be guarded against, what will governments decide.” The explosive growth of agreements (+328.0%) and sanctions (+283.0%) is the direct expression of that shift. In other words, the fall in tariff word frequency does not mean the tariff problem has been resolved; it means the bargaining has moved to a higher level.

Category U shows an internal split of its own: the concrete “risk” rose 85.2%, while the generalised “uncertainty” (неопределённость) is the only U family in the table to fall year on year (−28.5%, 98 mentions against 137). This indicates that perceptions of the threat are moving from the vague to the specific — no longer a general observation that “the outlook is unclear,” but the ability to name particular points of risk.

6. Long-Run Position (past two years)

Widening the view to the past eight quarters, the relative standing of the four countries’ f_ratios has undergone a fundamental reordering.

Table 10 Four-Country f_ratio Over the Past 8 Quarters

Source: CA-TPU Index Project.

The most striking feature is Uzbekistan’s sustained climb. Its f_ratio rose from 0.0222 in Q3 2024 to 0.0985 this quarter, roughly 3.4 times in two years, and the ascent has yet to peak. More important still is the crossover that occurred in Q4 2025: before then, Kazakhstan had held first place among the four for five consecutive quarters at levels of 0.098–0.121; from that quarter (Kazakhstan 0.0582 against Uzbekistan 0.0596) Uzbekistan moved ahead for the first time, and has widened its lead in each of the three quarters since. This change of leadership is the single most important development in the regional picture over the past two years.

Kazakhstan’s falling f_ratio needs careful interpretation. Its ratio dropped from a peak of 0.1209 to 0.0488, a decline of about 60%; yet as noted above, the absolute number of its AND hits rose rather than fell over the same period. What drove the ratio down was the abrupt expansion of the denominator — around Q4 2025 the volume of news collected for the country rose in a step change. Kazakhstan’s shift is therefore in essence a “denominator event,” not a “numerator event”: its absolute level of attention has not declined, it has simply been diluted in a much larger pool of news.

A third trend is the marked convergence of cross-country dispersion. Measured by the ratio of the highest to the lowest f_ratio among the four countries, dispersion peaked at 6.7 times in Q1 2025 (Kazakhstan 0.1209 against Uzbekistan 0.0181) and has fallen steadily since, converging to 3.2 times this quarter (Uzbekistan 0.0985 against Kyrgyzstan 0.0312); at the same time all four readings now fall within the 0.031–0.099 range. Trade policy uncertainty has thus evolved from a pattern of “high concentration in Kazakhstan alone” into a regional phenomenon covering all four countries — its reach is widening, not narrowing.

7. Trend Assessment

Drawing the quarter’s data together, four judgements are worth setting out.

First, the persistence of the uncertainty has exceeded earlier expectations. The quarterly average MA3 reached 163.93, an all-time high, and this is the only quarter in the index’s seventeen-year history in which all three months sat in the high-pressure zone (MA3 ≥ 150). By comparison, the second quarter of 2020, in the early COVID period, had a higher single-month peak but only two months above that line. The character of the present pressure is therefore not “more violent” but “more prolonged” — it applies pressure continuously at moderately high intensity rather than striking briefly with extreme readings. For firms this shape is the more costly one, because it denies any strategy of waiting for the storm to pass.

Second, the shape of the shock has changed from repeated oscillation to a single peak and fade. The first quarter had a twin-peak structure (surges in January and March); this quarter is a clean single-peak inverted V, with the weekly f_ratio declining almost monotonically from early April and ending the quarter at a low of 3.81%. Combined with the July data already published (raw index 95.16, MA3 143.32, with the high-pressure episode formally ending in that month), this quarter can be judged to be the top zone of the current pressure cycle and the starting point of its retreat. The focus of observation should shift from “will it go higher” to “how steep is the retreat, and at what level will it stabilise.”

Third, regional diffusion is complete and the driving core has shifted. Two years ago, trade policy uncertainty in Central Asia was largely a Kazakh phenomenon; today the f_ratios of all four countries lie within the same order of magnitude, dispersion has converged from 6.7 times to 3.2 times, and the position of highest concentration has passed from Kazakhstan to Uzbekistan. The implication is twofold. On one hand, policy changes in any single country now explain less of the regional composite, so risk assessment must weigh all four countries equally. On the other, Uzbekistan as the new driving core has an “agreement-led” agenda structure (its agreement family accounts for 67.9% of the regional total), which means its uncertainty stems more from rules under renegotiation than from shocks already delivered — a form of uncertainty typically accompanied by more frequent changes to contract terms.

Fourth, policy response has itself become the principal source of uncertainty. Category P keywords rose 135.9% year on year this quarter, the fastest of the three categories, with the government, ministries, law and strategy families generally up by more than 140%; at the same time the words that describe trade instruments directly — tariffs and duties — fell by more than 60% year on year. The contrast makes clear that the market’s focus is no longer quantifiable parameters such as tariff rates but the institutional responses governments will adopt, which are less predictable and transmit through longer chains. This is fully consistent with the 82.6% month-on-month surge of the law family in the July data, indicating that the trend is continuing.

Looking ahead to the third quarter, three variables will determine the course of the current cycle. First, whether the MA3, having fallen below 150 in July, continues to decline and stabilises near 100; if it does, the current high-pressure phase has formally entered its closing stage. Second, whether Uzbekistan’s f_ratio can retreat from its high of 0.0985: the country is the main fulcrum of the regional index at present, and its direction largely determines the regional level. Third, whether the warming of the law and tax families is sustained; if it is, the negotiations of the past two quarters are being converted into domestic institutional arrangements, and what firms face will shift from managing expectations to concrete compliance costs.

8. Data Appendix

Table 11 CA-TPU Index, Past 12 Months

Source: CA-TPU Index Project. Includes the month following the quarter (July 2026) for trend reference.

About the CA-TPU Index

The CA-TPU (Central Asia Trade Policy Uncertainty Index) was developed by the team of Professor Zhang Lijie at the School of Economics and Management, Xinjiang University, built on the economic-policy-uncertainty approach of Baker, Bloom and Davis (2016), covering the main Russian-language media of Kazakhstan, Uzbekistan, Kyrgyzstan and Tajikistan. The index measures the intensity with which these countries’ mainstream media simultaneously discuss the three topics of trade, policy and uncertainty; the base period is 2015–2019, normalised to a base-period mean of 100, and the national sub-indices are composited by trade weight (based on 2024 trade data, with Kazakhstan around 60% and Uzbekistan around one-third). The index is released monthly. Full methodology, historical data and trilingual reports are available at catpu.caiees.cn.

Data Analysis Method and Audit Support

The quantitative analysis in this report rests on eight sets of structured queries, covering the quarterly overview, the monthly path, the weekly trend, news totals, the cross-country comparison, month-by-month evolution within the quarter, media-source contributions and the long-run trend over the past eight quarters. In the keyword analysis, the various grammatical case forms of Russian words are merged and counted by root meaning; the merging rules are implemented as a deterministic mapping table, ensuring that the analysis is reproducible.

One point on data vintage should be noted. The CA-TPU index series (including the national sub-indices, the MA3 and the quarterly aggregates derived from them) follows a “frozen on publication” practice: figures for historical months are not revised retrospectively when the corpus is updated. Corpus diagnostics such as keyword frequencies, topic concentration and coverage ratios, by contrast, are recomputed in full on the latest corpus snapshot for each edition, so that comparisons across periods within a single edition are like-for-like. The diagnostic figures shown here for historical periods may therefore differ slightly from the values published in earlier monthly reports; this is a normal revision.

On the absolute level of Uzbekistan’s sub-index: since January 2026 the country’s readings have run persistently high, which is related to its sample characteristics within the 2015–2019 base period and is a methodological limitation the project team has identified. For this sub-index, attention should focus on direction and relative trend rather than on the absolute level.

To support third-party data audit and independent verification, this report is accompanied by the CA-TPU Q2 2026 Data Audit Package, containing the complete quarter-level aggregated datasets and the methodology documentation. Any third party can use it to reproduce independently all of the data analysis results presented in this report.

This report is a non-profit academic output, independently funded, and does not constitute investment advice or a basis for business decisions. Attribution: CA-TPU Index Project.