It follows that var(u)=var(8p)/(1-y2) as LeRoy and Porter noted. They base their volatility tests on our inequality (1) (which they call theorem 2) and an equality
restriction a2(p) +a2(8p)/(1-y2)=a2(p*) (theirtheorem 3). They found that, with postwar Standard and Poor earnings data, both relations were violated by sample statistics.