322 JoHN F. MUTH are compared in Figure 3. 1 for B= 2y and various values of fi. If current information is under-discounted (i= 1/2), the weight Vi attached to the latest observed price is very high. With over-discounting 1= 2),the weight for the first period is relatively low FIGURe 3. 1.Autoregression Coefficients of Expectations for Biased Use of Recent Information(o =Wn The model above can be interpreted in another way. Suppose that some of the firms have access to later information than the others. That is there is a lag of one period for some firms, which therefore form price expectations according to(3.8). The others, with a lag of two periods, can only use the following (320) ∑w Then the aggregate price expectations relation is the same as (3. 18), if fi represents the fraction of the firms having a lag of only one period in obtain ing market information(that is, the fraction of"insiders") 4. EFFECTS OF INVENTORY SPECULATION Some of the most interesting questions involve the economic effects of inventory storage and speculation. We can examine the effect by adjoining to(3. 1)an inventory demand equation depending on the difference between the expected future price and the current price. As we shall show, the322 JOHN F. MUTH are compared in Figure 3.1 for ,B 2y and various values of fi. If current information is under-discounted (f= 1/2), the weight VI attached to the latest observed price is very high. With over-discounting (fi 2), the weight for the first period is relatively low. UNDERDI COUNTING Vk .4, ~~~~~~~RECENT INFORMATION a 2 3 4 5 6 k .8 ~~~~~~~~UN31A5ED USE oF RECENT INFORMATION Uk * Io i i 2 4 5 1 OVERD SCOUNTINGC , ~~~~~~~~RECENT INFORMATIOW 1 2 3 4 5 6 FIGURE 3.1.-Autoregression Coefficients of Expectations for Biased Use of Recent Information. (wo = w, = ... == 1). The model above can be interpreted in another way. Suppose that some of the firms have access to later information than the others. That is, there is a lag of one period for some firms, which therefore form price expectations according to (3.8). The others, with a lag of two periods, can only use the following: , 00 (3.20) pt t-= Ew i=2 Then the aggregate price expectations relation is the same as (3.18), if fi represents the fraction of the firms having a lag of only one period in obtain￾ing market information (that is, the fraction of "insiders"). 4. EFFECTS OF INVENTORY SPECULATION Some of the most interesting questions involve the economic effects of inventory storage and speculation. We can examine the effect by adjoining to (3.1) an inventory demand equation depending on the difference between the expected future price and the current price. As we shall show, the