291 
terms show no significant effect of experience. The model 
that includes the interaction terms changes the estimates of 
the effect of both race and of experience considerably, 
making the main effect of experience highly significant and 
more than doubling the effect of race. Both the interac- 
tions with time are significant. Whites have more negative 
time dependency than blacks and inexperienced workers have 
more positive time dependency than experienced workers. 
Thus, young blacks have the most positive time dependency 
while experienced whites have the most negative. 
The effect of amount of unemployment compensation is 
significantly negative. The more compensation received, the 
longer the spell, if respondents receive any compensation at 
all. It is of considerable interest to note that the spline 
function fitted demonstrates that there appears to be a 
positive effect on the rate of reemployment of being covered 
by unemployment compensation. Thus for low levels of 
compensation there may be a higher rate of reemployment than 
for those that do not receive any compensation at all. This 
complicated relation between compensation and rates of 
reemployment suggests why there appears to be continuing 
controversy about the impact on compensation on duration of 
unemployment (Atkinson, Gomulka, and Micklewright, 1984). 
The effect of the average wage rate received in employment 
is positive, as predicted when this variable is interpreted 
as a measure of the persons resources. Whites and 
experienced workers have much higher rates of reemployment 
than blacks and inexperienced workers. 
Thus the frequent 
finding that young or black workers have short spells is 
largely a result of the positive time dependency. If these 
workers were in the same employment sectors as whites and 
experienced workers their spells would be considerably 
longer.