Several analytical models of the year-end inventory purchasing decisio
n of a LIFO firm have been developed (Cohen and Halperin 1980; Halperi
n 1979, 1981; Biddle and Martin 1985).1 This study finds empirical sup
port for the prediction of these models that, since only LIFO firms re
duce their tax burden by purchasing additional inventory (i.e., ''extr
a''inventory) at year-end, LIFO firms are more likely to purchase extr
a inventory at year-end than FIFO firms. This research also provides e
vidence that high-tax LIFO firms are more likely to purchase extra inv
entory at year-end than low-tax LIFO firms. This behavior is predicted
because a LIFO firms tax savings from purchasing extra inventory at y
ear-end are increasing in its marginal tax rate. Additional evidence o
f tax-motivated year-end inventory purchases is offered by tests which
find that (1) consistently high-tax LIFO firms accelerated their year
-end inventory acquisitions-and thus reduced their taxable income-to a
significant degree in the years immediately preceding the reduction i
n tax rates mandated by the Tax Reform Act of 1986; and (2) difference
s in tax status are not related to differences in fourth quarter inven
tory purchasing behavior for FIFO firms. The results of this study ind
icate that taxes have a sizable effect on the inventory purchasing pol
icy of LIFO firms. For example, (1) the estimated difference in the pe
rcentage of annual inventory purchases made in the fourth quarter betw
een high-tax and low-tax LIFO firms is equivalent, on average, to $12.
66 million of purchases (3.8 percent of ending inventory); and (2) the
estimated decrease in inventory purchases made in the fourth quarter
by LIFO firms that move from a high-tax status in one year to a low-ta
x status in the following year is equivalent, on average, to $28.89 mi
llion of purchases (12.1 percent of ending inventory). These large dol
lar amounts lend credence to the concern that year-end LIFO inventory
purchases made for tax purposes may lead to inventory management ineff
iciencies (Jannis et al. 1980, 186).