1. When does the market matter?
Baker, Stein and Wurgler(2003)

2. I. Introduction
Corporate investment and the stock market are positively correlated, in both the time series and the cross-section.
1. Traditional explanation: stock prices reflect the marginal product of capital (Tobin (1969)).
2. Irrationality explanation: Keynes (1936)
“Equity financing channel”: Stock prices contain an important element of irrationality, which results in that the effective cost of external equity diverges from the cost of other forms of capital. (Morck, Shleifer and Vishny (1990), Stein (1996) etc.)

3. I. Introduction Empirical tests
1. Traditional view: examine whether the stock market forecasts investment over and above other measures of the marginal product of capital mixed results
2. Measure inefficiency directly as the differences between market prices and a structural model of efficient prices, and then test whether investment is sensitive to this measure of inefficiency (Chirinko and Schaller 2001).
3. In this paper, the authors return to the theory to derive several cross-sectional predictions that are unique to a specific equity financing channel.

4. I. Introduction The financing channel from Stein (1996)
Those firms that are in need of external equity finance will have investment that is especially sensitive to the non-fundamental component of stock prices.
1. Build a proxy for the concept of equity dependence: “KZ index” based on the work of Kaplan and Zingales (1997).
2. Rank firms according to the proxy for equity dependence, and test in a variety of ways whether those that are classified as most likely to be equity dependent have the strongest correlation between stock prices (Q) and subsequent investment.

5. I. Introduction
Find strong support. Firms that rank in the top quintile of the sample in terms of the KZ index have a sensitivity of investment to Q that is almost three times as large as firms that rank in the bottom quintile.
In some specifications the investment of equity-dependent firms is more sensitive to Q than to cash flow.
3. While the finding is consistent with the equity financing channel, it also admits other interpretations. This is because Q contains more than just the non-fundamental component. It also embodies information

6. I. Introduction
about future profitability as well as as measurement error. As a result, these other components of Q may co-vary with the measure of equity dependence in such as way as to induce the pattern found in the investment-Q regression.
The authors uses “ future realized stock returns” (returns over the three years subsequent to the year in which investment is measured) as a noisy estimate of the future return expected by managers Expect that investment-returns sensitivity should be negative on average and increasingly negative in equity dependence.

7. I. Introduction
4. Financing patterns: find that firms with high values of the KZ index also have equity issuance that responds positively to Q and negatively to future returns.
Overall, the results offer support for a specific financing channel in corporate investment. They also complement other evidence that the cost of external equity has an important, independent effect on corporate financing and investment decisions (Baker and Wurgler 2000, 2002).

8. II. Hypotheses
Model
This paper uses a simplified version of Stein (1996) to develop several testable hypotheses.
1. First-best level:
2. Financing consideration: the equity may be mispriced by a percentage relative to the efficient-market value : the fractional debt capacity of the new assets. 廠商的最適化問題在p.974 eqs(1)~(3).

9. II. Hypotheses 3. Proposition 1: 若在負債限制下,可發行的新股可滿足最佳投資額的資金需求
(1) 若市場股價被高估, 或股價被低估但自有資金充足,則投資額會達到first-best;前者 ,後者e=0.
(2) 若股價被低估且自有資金不足,則投資額會低於first-best (underinvestment)
(2.1) 若 (where )
投資額為 , 發行新股
(2.2) 若 , 投資額 , e=0.

10. II. Hypotheses
4. When does the investment depend on the non-fundamental component of stock prices ? Two necessary conditions are as follows: the stock is undervalued, and the available wealth is so low that the firm would have to issue undervalued equity.
5. The sensitivity of investment to is governed by Firms with less debt capacity have investment that reacts more strongly to stock prices.

11. II. Hypotheses Testable hypotheses:根據定理一,發展出以下三個假說.
1. Hypothesis 1: define as firm as equity dependent if Equity-dependent firms display a more positive sensitivity of investment to Q than do non-equity dependent firms.
Q is taken to be an empirical proxy for .
It does not condition on whether firms are over- or undervalued.

12. II. Hypotheses
2. Hypothesis 2: Equity-dependent firms display a more negative sensitivity of investment to future stock returns than do non-equity-dependent firms.
Here “future stock returns” is used as a proxy for in stock prices (高估的股票未來會有低期望報酬,而低估的股票則傾向有高期望報酬).
The main impediment to testing Hypothesis 2 is that realized returns are likely to be a very noisy proxy for expected returns, and hence for mispricing.

13. II. Hypotheses
3. Hypothesis 3: Equity-dependent firms have equity issuance that is positively related to Q and negatively related to future stock returns.

14. III. Data
Unbalanced panel of Compustat firms that covers 1980 through 1999 (excludes financial firms, and firm-years with a book value under $10 million). The full sample includes 52,101 observations, for an average of 2,605 observations per year.
Investments: four measures of investment (panel A of Table I)
: capital expenditure/ book asset
: 加上研發支出
: 再加上銷管費用
: the percentage change in book value

15. III. Data
Financing: two measures of external financing activity (panel B of table 1)
: external equity issues to book assets (external equity issues: change in book equity minus the change in retained earnings)
: external finance (debt issues: the change in assets minus the change in book equity)
Equity-dependent firms:主要根據 Kaplan and Zingales (1997) (panel A of Table II)

16. III. Data
1. Five-variable KZ index for each firm-year as eq. 4. (當現金流量,股利與現金愈多時,資金受限的程度愈輕;當財務槓桿或Q越高,資金受限的程度越高)
2. Four-variable KZ index eq.5
在4式中,low dividend 與high Q皆可視為proxies for strong investment prospects.但Q also contains information about mispricing 故 for conceptual cleanness,作者於5式將Q去除.
3. Supplement the index with two other variables: AGEit :a proxy for reputation and access to lending market

17. III. Data :a proxy for industry debt capacity
Other investment determinants
year-fixed effects, firm fixed effects, Q, contemporaneous cash flow;
Q: (market value of equity+ assets- book value of equity ) / assets
and future stock returns (the raw three-year cumulated returns from CRSP, beginning at the end of the period in which investment is measured)

18. IV. Empirical results Hypothesis 1: investment and stock prices
1. Assign each firm to a quintile according to its median value of KZit over the full sample period. For each KZ quintile, estimate the investment equation eq Hypothesis 1 predicts that the coefficient b will generally increase as KZ increases (Table III: to 0.033).
2. Find no discernible pattern in c coefficient across the KZ quintiles (Table III: to 0.145).
3. Economic magnitudes:

19. IV. Empirical results
In the highest KZ quintile, the impact of a one-standard-deviation shock to Q(0.93)is to alter the ratio of capital expenditure to assets by 0.031(=0.033*0.93). This effect is substantial when compared to either the median or standard deviation of this investment measure(0.06, 0.079). (p.988)
4. Robustness of Hypothesis 1 results
Table IV 為表三基本結果的robust test.
(1) Row 1: baseline specification
(2) Row 2: five-variable version of KZ index (including Q)

20. IV. Empirical results
(3) Row 3: four-variable version of KZ index, but classify firms based on their 5-year median values of KZ
(4) Row 4: reclassify firms every year; assign a firm to a KZ quintile in any given year t based on its value of the index in year t-2
(5) Row 5: scale the elements of KZ index by PP&E
(6) Row 6: add the lagged value of the capital expenditure ratio as the explanatory variable
(7) Row 7: delete the cash flow term from the regression

21. IV. Empirical results
(8) Row 8: subsample similar to Kaplan and Zingales(1977) ( observations 26,725)
(9) Row 9: the complementary subsample
(10)Row 10: the results are not overly sensitive to the exact weights that the KZ index gives to its four components
5. Table V: 以其他的投資測度來檢驗
In a compact fashion, run eq.7. Iit代表四個投資測度
As predicted, the coefficient c is significantly positive for each of the four investment measures.
The competing explanation

22. IV. Empirical results
6. Decomposing KZ and the effect of other indicators
直接將KZ index 中的變數作為解釋變數,觀察個別變數如何影響投資的敏感性,如eq.8,其中c1~c3 應為負, c4 應為正.
Panel A of Table VI: the predictions are largely borne out. The sharpest results are for c4 .
Panel B of Table VI: 考慮額外的兩個變數 firm age and industry cash flow volatility. Re-estimate eq.8 augmented with the interactions of these two variables with Q (predict negative c5 and positive c6 )

23. IV. Empirical results
Hypothesis 2: Investment and future stock returns. Replace Q in the investment equation with future stock returns. See eq.9. b is expected to be negative on average whether variation in R reflects mispricing or differences in the rational cost of capital. The unique prediction is that b should be more negative for equity-dependent firms.
1. Measurement error problem: 利用Vuolteenaho (2002)的方法計算signal-to-noise ratio for each of the KZ categories, the differences are very small.

24. IV. Empirical results
2. Table VII: the basic prediction of Hypothesis 2 is supported. The coefficients on future returns are negative and significant in each KZ quintile
( to ).
Hypothesis 3: financing, stock prices and stock returns. The equity issuance of equity dependent firms is positively related to Q and negatively related to future returns. 前述迴歸的被解釋變數改為融資變數.
1. Panel A of Table VIII: financing and Q

25. IV. Empirical results
Row 1: the sensitivity of equity issuance to Q is strongly significant across all KZ quintiles (0.021 to 0.064). Thus firms classified as most likely to be equity-dependent have equity issuance that is strongly tied to their stock prices.
Row 2: According to Proposition 1, equity-dependent firms should also be raising a significant amount of debt on the margin. For firms in KZ quintile 5, the sensitivity of total finance to Q is 0.136, whereas for equity issuance alone is

26. IV. Empirical results
2. Panel B of Table VIII: changes the proxy for mispricing from Q to future returns. Equity issuance is significantly negatively related to future returns for high-KZ firms, and the response of equity issuance to future returns in this group accounts fro roughly half of the response of total external finance to future returns.