A Bad Year, or a Bad Place?

Start with a simple question

When crop yields differ, is it mainly because of where the crop was grown or when it was grown?

This method splits the differences into four parts: crop, country, year, and everything left unexplained.

First test: mix all crops

Put every crop into one model and ask what explains the differences.

Where yield differences come from

Mixing all crops gives the wrong picture

Crop type looks most important only because crops have very different yield levels. When each crop is studied separately, country becomes the largest source of difference.

All crops together

73.9%15.9%

Crop 73.9%Country 15.9%Unexplained 8.9%Year 1.3%

Each crop tested separately

69.1%23.9%

Country 69.1%Unexplained 23.9%Year 5.6%

Evidence detail

Test	Country	Crop	Year	Unexplained
All crops together	15.9%	73.9%	1.3%	8.9%
Each crop separately	69.1%	not included	5.6%	23.9%

Why this answer is misleading

The result says crop type explains 73.9% of the difference. But this is not very useful.

Sugarcane and wheat have very different yield levels. Mixing them mostly confirms that sugarcane is not wheat.

Better test: study each crop separately

Now ask the same question one crop at a time: does place or year matter more?

One crop at a time

For most crops, place explains far more than year

Tea is the main exception. Its yields rose steadily over 63 years, so year matters more than country.

Tea leaves

57.9%

Soya beans

15.0%

Wheat

14.1%

Maize (corn)

12.3%

Potatoes

11.9%

Rice

8.8%

Barley

8.3%

Sorghum

3.0%

Taro

0.8%

CountryYear. The number on the right is the year share.

Evidence detail

Crop	Country share	Year share	Unexplained	Countries
Tea leaves	25.7%	57.9%	16.4%	46
Soya beans	63.1%	15.0%	21.9%	43
Wheat	69.6%	14.1%	16.3%	84
Maize (corn)	69.1%	12.3%	18.6%	126
Potatoes	70.8%	11.9%	17.3%	110
Rice	71.6%	8.8%	19.5%	98
Barley	77.3%	8.3%	14.5%	67
Sorghum	80.0%	3.0%	16.9%	74
Taro	85.6%	0.8%	13.6%	39

Across 63 crops, country explains a typical 69.1% of the difference. Year explains only 5.6%. Place matters about twelve times more.

Tea is different. Year explains 57.9% because tea yields rose steadily in many countries.

When this is useful

Use this method when a result has several possible sources and you need to know where to look next:

• Country is largest: check farming methods, climate, irrigation, and local policy.
• Year is largest: check weather, new crop varieties, and changing technology.
• The unexplained part is largest: an important factor may be missing from the data.

What this test cannot tell us

Year does not mean weather alone. It also includes long-term changes such as better crop varieties and farming inputs.

The dataset includes only countries and crops with records for every year from 1961 to 2023. This makes comparison easier, but leaves out incomplete records.

This test does not prove cause. “Country” combines soil, climate, irrigation, crop varieties, and policy. The result shows where differences appear, not exactly what caused them.