Two B-School economics professors, Andrew Bernard and Meghan Busse, constructed an economic model of Olympic medal-winning that was recently published in the Review of Economics and Statistics. Since there's no better way to test a model than by forecasting, the authors put on the Web their forecast for this year's Games.

How did they do?

Below I list their forecasts for the 34 countries on their Web page and the actual result. In a third column, "naive forecast", are listed the number of medals won by each country in the last Summer Olympics.

The mean square error of Bernard and Busse: 28.5.
The mean square error of the naive, no-change forecast (not including the last three nations): 25.6.

Oh, well . . .

(To be fair, their model missed badly on just two countries, Greece and Japan, and they thought Greece would benefit a lot from being the host. Without Greece, the naive model's MSE is 25.4 and their model's MSE is 24.9. And they also forecast just the number of gold medals. There, even with Greece included, their model outperforms no-change by 9.6 to 10.9)





Naïve Model Bernard & Busse Actual
U.S.
97
93
103
Russia Fed.
88
83
92
China
59
57
63
Germany
57
55
48
Australia
58
54
49
France
38
37
33
Italy
34
33
32
U.K.
28
27
30
Greece
13
27
16
South Korea
28
27
30
Cuba
29
25
27
Romania
26
23
19
Netherlands
25
21
22
Ukraine
23
20
23
Japan
18
19
37
Hungary
17
14
17
Belarus
17
13
15
Canada
14
13
12
Poland
14
12
10
Brazil
12
12
10
Spain
11
11
19
Sweden
12
11
7
Bulgaria
13
10
12
Norway
10
8
6
Switzerland
9
8
5
Czech Rep.
8
6
8
Mexico
6
6
4
Indonesia
6
6
4
Ethiopia
8
5
7
Kazakhstan
7
5
8
Denmark
6
5
8
Kenya
7
4
7
Jamaica
7
3
5
Georgia
6
3
4
Turkey
4
 
10
Poland
14
 
10
Thailand
3
 
8

Comments