LQ45 tariff tests leave out the 8 April reopening

The session a trader would most want to examine is missing from this event study's hypothesis tests.

Nareswari and Kusuma examine the response of Indonesia's 45 LQ45 stocks to Trump's reciprocal tariff. The paper describes a levy of up to 32% on partner countries including Indonesia. Announced in the 2 April 2025 Liberation Day speech, the news reached Jakarta on 3 April. But the Indonesia Stock Exchange was shut for Idulfitri from 28 March to 7 April. Prices could first respond on 8 April, the authors' event day, t0. The authors identify that closure as the study's methodological distinction: news could accumulate before trading resumed. We see that framing as the paper's real contribution. Their hypothesis tests nevertheless exclude t0 from both windows.

Expected returns follow a standard market model against the IHSG, fitted over 100 trading days (t-105 to t-6). Abnormal return (AR) is the difference between the actual and expected return. Trading Volume Activity (TVA) divides shares traded by shares outstanding. The authors average each measure for each stock over t-5 to t-1 and t+1 to t+5. All four series fail Shapiro-Wilk. They then compare the windows using a paired Wilcoxon signed-rank test, finding a significant AR difference (Z = -2.918, p = 0.004). TVA shows no significant difference (Z = -1.428, p = 0.153).

The authors interpret these results as evidence that the tariff conveyed negative information. They also conclude that LQ45 absorbed it rapidly "without post-announcement drift" and that the exchange is semi-strong efficient. A trader would read no drift as no remaining return to harvest after the reopening print.

What does 8 April capture?

Both the announcement and the reopening sit on that date, and the test drops the session. With t0 excluded, the authors compare five sessions before 28 March with the five sessions after 8 April. Neither window contains the reopening price, which incorporates news from the entire closure.

Later news enters the post window as well. The US 90-day tariff pause on 9 April 2025 falls within t+1 to t+5; we did not find a discussion of it in the paper. The test thus compares a late-March week with a mid-April week containing a policy reversal.

Thirty-five of 45 stocks improved, yet mean abnormal return fell

The direction of the significant AR result clashes with the authors' interpretation. Post-minus-pre ranks are positive for 35 stocks (rank sum 776) and negative for 10 (rank sum 259). There are no ties. For the typical LQ45 stock, market-adjusted return was higher after reopening than before the holiday. That majority drives the Wilcoxon statistic.

The means supply the negative reading. Mean AR falls from -0.0106 to -0.0141, while its standard deviation almost triples from 0.0430 to 0.1179. The minimum after the event is -0.7207, compared with -0.2112 beforehand. In a 45-stock mean, that observation contributes -0.016. Its contribution exceeds four times the -0.0035 difference between the means. An average abnormal return of -72% for an LQ45 constituent looks more like an unadjusted corporate action or a data error than a tariff response. We did not find any examination of it.

The authors say the preponderance of positive ranks demonstrates a market response through price adjustment. They do not explain the direction. There is an arithmetic error too: 776/35 equals 22.17, whereas the table reports 22.71.

The choice of benchmark adds another difficulty. According to the paper, LQ45 represents more than 50% to 60% of IDX capitalisation. Using the IHSG in a market model strips out most of a market-wide macro shock. What remains measures relative performance against an index largely composed of these same stocks.

The abstract misreports volume

The abstract claims aggregate tests find significant differences in both AR and TVA. Yet Table 3 records H2 as rejected at p = 0.153. The conclusion follows the table, explaining the absence of a volume spike by LQ45's already high liquidity. The authors also explain the null result as a volume spike concentrated around the announcement day. That day is t0, precisely the session excluded from their test. Even so, both the abstract and the discussion continue to use turnover as evidence.

The descriptive account contains another error: mean TVA declines from 0.000165 to 0.000123, which the paper describes as a slight increase. Again, the ranks tell a different story, with higher post-event trading for 29 of 45 stocks. So 29 of 45 stocks rose while the cross-sectional mean declined, and the Wilcoxon test detects no significant change (p = 0.153).

Normality reporting also needs care. The table gives TVA-post Shapiro-Wilk as Sig. < 0.001; the text gives 0.002. And although the method section says it tests differences, the table tests levels.

Pricing speed remains unmeasured

The paper supplies three tables covering descriptives, normality and the Wilcoxon test. We found neither a daily abnormal return series nor a cumulative path. Two five-day averages cannot distinguish a one-day adjustment from a move unfolding across t+1 to t+5. The description of a "short-lived" reaction and the claim of no drift are therefore assertions. The semi-strong efficiency conclusion depends on two paired Wilcoxon tests across 45 stocks, for a single event, excluding the event day. The TVA test (p = 0.153) is insignificant.

We could not run this ourselves.

We lack Indonesian daily prices, volumes and shares outstanding for the LQ45 constituents, as well as the IHSG. A substitution using US equities would remove the eleven-day closure, the feature that makes this setting unusual.

A daily table covering t0 through t+5, including the reopening, would change our view if it showed cumulative abnormal returns flattening after day one. That result would also need to survive scrutiny of the -0.7207 observation. Until such a table exists, the paper establishes a cross-sectional shift whose sign remains uncertain. The speed at which the tariff entered prices remains unshown.