i am taking some assumptions here : that the preccession of earth does not change the ecliptic w.r.to earth's equator. then:
the perpendicular movement MUST be due to the relative velocity of the star w.r.to sun.
the position of star w.r.to distant stars as seen form earth ( for a particular time on a sidereal day, plotted day after day , all at a particular place ) will be like this :
component parallel to ecliptic moves sinusoidal with time ( with an amplitude of .5 arcsec and time period of 1 year )
component perpendicular will move linear with time ( speed : 1 arcsec/year )
overall, it moves like a sine wave.
( y axis : line of ecliptic )