Corrections made in the seventh printing.

These corrections were sent to the publishers in the early spring of
1999, and the seventh printing was produced during the late spring of
that year.

Corrections are listed by the name of the ps and pdf files, which
is based on the page number(s) affected.

pg052 The subscript of the argument of x_2 in the second line after the
      unnumbered equation has been corrected.

pg066 Typo corrected in last paragraph. "models and." ==> "models."


pg074 The strict inequality was changed to a non-strict one in equation
      (3.14), there was a minor change just after it, and there were
      two minor changes in the paragraph following that equation.

pg091 An expression has been transposed in the sentence before (3.48). In
      the fourth printing, a sign error was corrected (verbally) in the
      sentence after (3.45).

pg096-7 On page 96, the locations of the parentheses in the equation just
      above (3.58) have been changed. On page 97, a minor typo has been
      fixed in the last full paragraph: "uncertainly" ==> "uncertainty"

pg118 t_t should be tau_t in the last paragraph of Section 4.4.

pg126 The statement of Theorem 4.2 was not quite correct.

pg132 The dimensions of Y and Z in Definitions 4.11 and 4.12 were too
      small by one.

pg135 The argument of the summation in the first displayed equation
      should be (y_t - \mu) not y_t (fixed in seventh printing)

      The factor of n^{-1} in the denominator of the last display should
      not have been there and was removed.

pg160 The equation number was in the wrong place in (5.47).

pg164 The second term in the middle of the first equation has been
      transposed.

pg169 In the fifth printing, n^{-1/2} was replaced by n^{1/2} on the
      left-hand side of equation (5.71). In the seventh, one set of
      parentheses was removed in the third line of the unnumbered
      equations between (5.70) and (5.71). What was there previously was
      not wrong, but it was inconsistent with the discussion following
      the equations.

pg173 A missing transpose has been added on the left-hand side of
      equation (5.78).

pg188 Some unnecessary parentheses were removed three lines after (6.18).

pg215 Typo corrected in last paragraph of Section 7.3. "many" ==> "may"

pg221 The discussion of the relationship between the second-stage
      residuals and the true 2SLS residuals, following (7.30), was
      wrong and has been corrected.

pg230 The discussion leading up to (7.48) was changed slightly, to accord
      with the correction on page 221. In the second printing, "="
      was replaced by "\equiv" in (7.45) and (7.46).

pg232 "n-k+r" ==> "n-k" in the second-last sentence of the last paragraph
      in Section 7.7.

pg264-5 Subscripts have been added to the symbols for the Hessian and
      Information matrix in the last display on page 264. On page 265,
      the middle term in (8.47), which was incorrect, has been dropped.
      The sentence of which (8.47) is part has been shortened, and the
      following sentence has been modified.

pg276 An incorrect (but harmless) minus sign has been removed in the
      second of the two unnumbered equations immediately above (8.73).

pg285 The first factor in the second term of (8.93) should be n/2.
      There have also been two minor but consequential changes in the
      fourth sentence of the middle paragraph.

pg302 Incorrect factor of -2 removed from expression following (9.32).
      A minor correction to (9.31) was made in the third printing; the
      zero in (9.31) is now in boldface because it is actually a vector
      of zeros.

pg314 Extra transpose removed from expression (9.61).

pg319 An error was corrected in the expression for the matrix of
      derivatives near the bottom of the page.

pg348 A serious error in the only equation was corrected in the second
      printing. In the seventh, the year of the Poirier reference was
      corrected.

pg353 In the last paragraph, "case in which \alpha = 1" should be
      "case in which |\alpha| = 1".

pg360 In third line, max(p,q) replaces p+q. Also, footnote number changed.

pg362 In the second printing, 1.778 was changed to 1.771 in eighth line
      from the bottom. In the seventh printing, the footnote number was
      changed.

pg369-70 In several places on pages 369 and 370, the arguments of x_t'
      were changed in the third printing. Original version omitted \rho
      as an argument. A missing tilde was added near the bottom of page
      370 in the fourth printing. In the seventh printing, the line
      before (10.100) was corrected, the first reference to the GNR
      (7.38) was corrected in the paragraph containing (10.103), and the
      second line following (10.100) was corrected.

pg385-86 In the third printing, in (11.22), boldface "b" replaced "beta",
      and the first term on the right-hand side of the second line of the
      second display on page 385 was corrected. Also, on page 386,
      "quantity" was changed to "expectation of" in the first line of the
      second full paragraph.
 
      In the seventh printing, the second term on the right-hand side of
      the second line of the second display on page 385 was corrected, and
      boldface y was changed to boldface u in the sixth line on page 386.

pg391 The reference to (11.35) just before (11.37) has been changed to
      avoid the implication that (11.35) is an equation, which it is not.

pg430 The same typo was corrected twice in the first full paragraph.
      "(12.36)" ==> "(12.35)"

pg461 There were changes on this page in the fifth printing. In the
      seventh, 2n was replaced by -2n following (13.57).

pg470 Near the end of the long, middle paragraph max(p,q) replaces p+q.

pg472 The "G" in the first display should not have been bold.

pg484 Minor change to Poirier reference.

pg506 In the second sentence following (14.45), "added to" has been
      changed to "subtracted from".

pg543 An error was corrected in the second term in (15.55). The argument of
      phi should be divided by sigma. In addition, the left-hand
      side of the equation before (15.55) was changed to make it clear
      that the probability is conditional on y^*_t.

pg558 There are some substantive changes in the paragraph following the
      one that contains (16.22).

pg577 The equation number for (16.64) is now in the right position.

pg593 The 0 in the last line of Theorem 17.2 is now in boldface, since it
      is a vector.

pg619 Just before (17.75), the matrix D was partitioned in the wrong way.

pg636 The plus signs have been changed to minus signs in the two equations
      at the top of the page.

pg642 The order of "restricted" and "unrestricted" was inverted following
      (18.33).

pg664-5 Expression (18.81) has been corrected. There should be only one
        summation, not two. Expression (18.83) has received even more
        extensive corrections, and the sentence following it has been
        changed. There has also been a change of wording in the first
        full paragraph on page 664.

pg700-1 The correct section is now referred to in the last line of page
        700, and the argument of the polynomial equation is now correct in
        the third line of page 701.

pg716 The zero in equation (20.20) is now in boldface.

pg821 Several references were in an incorrect format. Also, references
      were updated in the second printing.

pg842 Poirier (1978a) formerly had the wrong year. Poirier (1978b) has
      therefore had the b added.