How to interpret beta in VECM? After running the Johansen test for cointegration, 1 cointegrating vector was found. Let's say the following cointegrating equation was estimated (A was normalized)
A              B            C
1.000000    -0.335218   -0.63412

A 1% increase in the A leads to a 1.34% increase in the B in the
long run. Is this the correct interpretation? 
 A: Your interpretation is not correct. By only knowing the cointegrating vector you cannot be sure which of the series influence which other series, and what the effect sizes are. For example, B might influence A but not necessarily. To find out the effect of B on A, you would have to look at the estimated coefficients* in the VECM rather than the coefficients of the cointegrating vector (which technically are estimated before estimating the VECM). 
(Your interpretation would be more related to the loading coefficients $\alpha$ than the cointegrating vector $\beta$, but even there the interpretation would differ a bit from your statement. Actually, if the coefficient vector above was the loadings $\alpha$, you could say that a unit increase in B leads to a -0.33 units decrease in A in the next time period.)
What the coefficients of the cointegrating vector mean is just what relative weights of the series are needed to produce a stationary combination from the original (integrated) variables.

* By estimated coefficients I mean the loading matrix $\alpha$ that premultiplies the matrix of cointegrating vectors (in your case there is only one vector, so $\alpha$ is a vector) and the $\Gamma$ matrices that multiply the lagged values of the first-differenced variables. 
