Because at any given moment in time there's likely no such thing as a par 10y German bond. For example, the current 10y benchmark is the Aug 24 bund. As you can see, it's more like a 9y 10mo bond, rather than a 10y bond. In order to calculate the 10y par yield from the specific issue yields in the mkt, you need to interpolate, which is what what various models and methodologies do. If you look through the BuBa data, you'll find another data set (not Svensson). I honestly don't remember what the differences between the two methods are. You should read the notes and decide which one works better for you.
well, thank you. I've read a lot earlier about spot, instantaneous forward etc. But it still hard to understand which one is more efficient to use.
If you want to understand this stuff better, seek ye the series of papers titled "Understanding the Yield Curve" by Antti Ilmanen...
While you are at it, read the book by the same author "Expected Returns" http://www.amazon.com/Expected-Returns-Investors-Harvesting-Rewards-ebook/dp/B004YK0JLW/ref=sr_1_1?s=books&ie=UTF8&qid=1411644804&sr=1-1&keywords=expected returns It is easily one of the best book of the century