Shwartz (2003) writes:
Two studies with students have assessed the test-retest reliability of
the ten values, as measured by the PVQ. Respondents completed the PVQ
twice, separated by an interval of two-weeks in Israel and 6 weeks in
Germany. The test-retest reliabilities (Israel & Germany) were
moderate to high: power .84 & .77, security .88 & .70, conformity .86
& .72, tradition .81 & .80, benevolence .82 & .62, universalism .83 &
.75, self-direction .66 & .70, stimulation .74 & .76, hedonism .84 &
.65, achievement .83 & .82.
However, It would be good have moreinformation in that description to really evaluate the quality of the data. I.e., Was the data ipsatised? What was the sample size?
Schwartz, Caprara, and Vecchione (2010) measured basic values using the PVQ in a sample 1030 participants before and after an election. They report that
Test-retest correlations ranged from .65 (benevolence) to .75 (achievement and hedonism).
This 2010 paper seems more relevant given the large adult sample.
Overall, these test-retest correlations are a little lower than what I've often seen with a well-validated measure of personality, where you might see test-retest correlations more in the .75 to .85 range.
In general, it's difficult to say whether these slightly lower correlations reflect the slightly more volatile nature of values or whether it is due to some measurement issues. In particular, I imagine having 3 items for some scales should lead to slightly lower test-retest correlations.
Schwartz, S. H. (2003). A proposal for measuring value orientations across nations. Questionnaire Package of the European Social Survey, 259-290.