Step 2 · Lag explorer

Which climate signal leads which price, and by how long?

Weather shocks take time to reach a price: a drought has to hit a harvest, the harvest has to come in short, and traders have to notice. So the first question is about timing. Each cell below is the correlation between a commodity’s monthly price change and a climate index some months earlier.

Correlations this small (|r| around 0.1) are typical for monthly commodity returns. They are a screen, not a forecast test: the backtests ask whether any of these leads survive an honest, out-of-sample comparison.

Commodity

Prices

Period

Wheat HRW: correlation of each month's price change with each index, 0 to 24 months earlier

Whole record, nominal log-returns. Click a cell to inspect it; arrow keys move, Enter selects.

Index up, price downIndex up, price up(-0.15 … 0.15)

outside the 95% no-association band

Show the numbers

Correlation by index and lag
Lag	MEI.v2	ONI	AO	PNA	SAM	DMI
0	-0.100	-0.065	0.040	-0.035	0.001	0.026
1	-0.091	-0.054	-0.022	0.026	0.006	0.027
2	-0.089	-0.036	-0.028	0.018	-0.011	-0.009
3	-0.094	-0.010	0.014	-0.002	0.016	0.006
4	-0.048	0.023	-0.011	-0.035	0.000	0.025
5	0.001	0.056	0.018	-0.005	0.024	0.044
6	0.009	0.085	0.046	0.032	-0.005	0.054
7	0.034	0.102	-0.001	0.074	0.026	0.057
8	0.065	0.112	0.006	0.027	-0.028	0.101
9	0.071	0.116	-0.016	0.036	-0.037	0.106
10	0.077	0.122	0.006	-0.015	-0.015	0.080
11	0.110	0.126	-0.021	-0.051	-0.033	0.088
12	0.119	0.117	-0.010	0.024	0.008	0.086
13	0.115	0.097	-0.035	0.010	-0.005	0.083
14	0.095	0.075	-0.047	0.079	-0.023	0.051
15	0.079	0.057	-0.033	-0.007	-0.016	0.008
16	0.056	0.038	0.000	-0.052	0.032	-0.016
17	0.031	0.017	-0.016	-0.001	-0.039	-0.006
18	0.018	-0.006	-0.011	0.004	-0.129	0.002
19	0.009	-0.022	0.019	-0.053	-0.063	0.020
20	-0.011	-0.033	0.008	-0.002	0.041	0.026
21	-0.003	-0.037	-0.029	0.013	0.025	-0.035
22	0.000	-0.035	-0.010	-0.019	0.082	-0.071
23	0.000	-0.028	0.037	0.002	0.070	-0.089
24	0.018	-0.019	0.037	-0.099	0.067	-0.062

29 of 150 cells clear the 95% band; about 8 would by chance alone even if no index had any link to prices. Treat isolated dots as leads to test, not findings. Lag 0 (same month) is shown for reference but cannot be used to forecast.

ONI against Wheat HRW, by lag

Grey wash: the 95% band expected with no association (roughly ±1.96/√n).

Each month, 11 months apart

r = +0.126 over 790 months (recomputed here from the raw series; the database says +0.126).

Oceanic Niño Index

Three-month running mean of sea-surface temperature anomalies in the Niño 3.4 region. Re-labelled by the last month of each season so no future month leaks into a forecast.

Source: NOAA CPC (via PSL), Jan 1960 to Sep 2026. Raw file