# Value of conglomerates and capital market conditions.

This article studies variations in the value of diversification across time under various capital market conditions. I find that when external capital is more costly at the aggregate level the value of conglomerates increases relative to focused firms. I also find that this increase is greater for financially constrained conglomerates, such as bank-dependent or small conglomerates. My findings support the theories on the advantage of diversification over focus. They suggest that the ability to substitute external capital markets with internal capital markets creates value for conglomerates when the financing cost in external markets is high, especially for those conglomerates that are financially constrained.In this article, I study variations in diversification value over time under various market conditions, and how conglomerate values respond to shocks in external capital markets.

Market macro conditions, especially the overall ease of external financing, can affect conglomerates and focused firms differently. When capital markets are constrained at the aggregate level, both conglomerates and focused firms face limited access to external capital. However, during such market conditions, conglomerates have an advantage over focused firms, in that conglomerates can substitute their internal markets for the depressed external markets and thus better survive economic and financial hardships.

Many studies on internal capital markets suggest that internal capital allocation can create value for conglomerates by avoiding costly external markets (see, e.g., Bhide, 1990, Stein, 1997, and Matsusaka and Nanda, 2002). These studies imply that the value of internal capital markets should depend on external market conditions. When the cost of external funds increases, internal capital allocation should become more valuable to conglomerates, especially for conglomerates that face costly financing constraints in external capital markets. However, to my knowledge, no empirical study has tested this implication. My article intends to fill this gap.

I test two hypotheses in the article: first, when external capital is more costly at the aggregate level, the value of conglomerates increases relative to the value of focused firms; and second, such value increase is greater for those conglomerates that are more financially constrained in external capital markets. I use bank-dependency to proxy for the financial constraint faced by each firm, assuming that bank-independent firms face less external financial constraints than bank-dependent firms. Thus, my second hypothesis essentially is that bank-dependent conglomerates experience greater value increases than bank-independent conglomerates when external financing cost increases at the aggregate level.

I test the two hypotheses at both the macro level and the firm level. In the macro-level study, I use various conglomerate return indexes to capture the value difference between a portfolio of conglomerates and a portfolio of focused firms. In the firm-level study, I follow the method used in the diversification discount literature and use conglomerate excess value to measure the value of a conglomerate relative to its focused peers.

I find that during my sample period, 1984 to 1997, the value of conglomerates relative to focused firms (i.e., the value of diversification) gradually declined. This downward trend of the value of diversification could be attributed to the development of financial markets, which has increased the efficiency of external capital markets and comparatively reduced the cost advantage of internal capital markets. I also find that conglomerates, relative to focused firms, are valued more favorably when external capital is more expensive, e.g., when interest rates are high or when equity is undervalued. These findings support my first hypothesis. They suggest that internal capital allocation functions as a valuable substitute for external capital markets, especially when external financing becomes more costly.

I then disaggregate the full sample of conglomerates and focused firms into two subsamples and compare bank-dependent firms with bank-independent firms. I find that the value of bank-dependent conglomerates declines over time, but that the downward trend is less significant for bank-independent conglomerates. I also find that when external financing costs increase at the aggregate level, the value of bank-dependent conglomerates increases relative to their focused peers. In contrast, the relative value of bank-independent conglomerates is less sensitive to the change in external financing costs. These findings are consistent with my second hypothesis. They suggest that the availability of internal capital markets to substitute for external markets is more valuable to financially constrained conglomerates than to conglomerates that are not financially constrained.

Finally, I also check the robustness of my regression results. In the first robustness check, I consider the possibility of a spurious relation given that both my measures of conglomerate value and capital market conditions could be non-stationary. I calculate the first difference of both measures to create stationary variables and find similar results from the regressions that use these differenced variables. In the second robustness check, I use firm size and the Kaplan and Zingales 0997) index (KZ index) as alternative measures of firms' financial constraint (instead of bank-dependency). Small firms face more limited access to external financing in tightened capital markets than large firms do; and firms with high KZ are more financially constrained than firms with low KZ. The results based on these alternative measures are similar to those based on bank-dependency. In general, conglomerates that are financially constrained, such as small conglomerates or conglomerates with high KZ, experience value increases relative to their focused peers in the case of a negative shock in external capital markets. Conglomerates that are not financially unconstrained, such as large conglomerates or conglomerates with low KZ, do not experience such value increases.

My article contributes to the empirical literature on the difference between conglomerates and focused firms. There are two broad approaches in this literature. The first approach focuses on the capital allocation in internal capital markets. Many studies suggest that internal allocations are inefficient at the average level due to cross-subsidization in internal capital markets (see, e.g., Lamont, 1997, and Shin and Stulz, 1998). However, Chevalier (2004) points out that some of the cross-subsidization results in these studies could be attributable to selection bias. Thus, the conclusion on the efficiency of internal capital markets is still ambiguous. (1) The second approach is to compare the overall value of diversified and focused firms with discount measures (Lang and Stulz, 1994, Berger and Ofek, 1995, Servaes, 1996, Lins and Servaes, 2002, Doukas and Ozgur, 2004, and Jandik and Makhija, 2005) or event returns (Matsusaka, 1993, and Hubbard and Palia, 1999). My article uses the latter approach. The findings in the article suggest that the value of conglomerates' internal capital markets depends on the cost of financing in external capital markets.

The article is organized as follows. Section I discusses my sample selection and the construction of variables. Section II presents evidence on the relation between the value of conglomerates and capital market conditions. Section III checks robustness and Section 1V concludes.

I. Sample Selection and Variable Constructions

In this section, I discuss my sample of conglomerates and focused firms and the variables used in my empirical tests.

A. Sample Selection

I obtain my sample from the Compustat Industry Annual database and the Compustat Segment database. From 1976 to 1997, the Statement of Financial Accounting Standards 14 (SFAS 14) required public firms to disclose information on segments that are significant to the firms. Typically, a business segment is considered significant if its revenues, assets, or profits account for more than 10% of firm totals. In 1997, SFAS 131 ruling changed the requirements for segment reporting. The new ruling requires firms to report segment information based on operating segments. Compustat is in the process of restating historic segment data to comply with the new ruling. However, the restated data go back only to 1984 at this time. Thus, my sample period covers from 1984 to 1997.

I impose the following restrictions on the sample. I exclude firms that operate or have segments in the financial services industry (SIC codes 6000-6999) or utilities (SIC codes 4900-4999). I also exclude firms with sales less than $20 million and firms that are missing segment values for sales or primary SIC codes.

I classify the sample into focused and diversified firms (i.e., conglomerates). In a given year, I classify a firm as a focused firm if the firm reports only one segment or if all the firm's segments share the same primary 4-digit SIC code at both the beginning and end of the year. I classify a firm in a given year as a diversified firm or a conglomerate if the firm has two or more segments with different primary 4-digit SIC codes throughout the year. (2) Thus, my sample in each year excludes both diversifying firms (i.e., firms diversifying from focused firms to conglomerates in that particular year) and refocusing firms (i.e., firms that refocus from conglomerates to focused firms in that particular year).

By excluding both diversifying and refocusing firms from the sample, I can ensure that the difference between diversifying and refocusing firms does not dictate the time-series variation in the value of conglomerates. Campa and Kedia (2002) suggest that the variation in the average excess value over time could be driven by the difference between the entering and the exiting single-segment firms in the sample. Similarly, the exit (refocusing) and the entry (diversifying) of conglomerates in my sample of conglomerates might bias my results as well, since it is possible that diversifying firms are better performers and refocusing firms are worse performers. In a robustness check that I do not present in this article, I also include both diversifying and refocusing firms in my sample and obtain results similar to those presented here.

My final sample consists of 7,836 firms with 42,669 annual observations, of which 6,576 focused firms account for 32,796 observations and 1,973 diversified firms account for 9,873 observations. Note that the number of focused firms and the number of conglomerates do not add up to the total number of firms. This difference is because some firms changed from conglomerates to focused firms or from focused firms to conglomerates during my sample period. Panel A of Table I shows the yearly distribution of the sample. Some of my empirical tests do not include the full sample due to incomplete information on certain financial variables (e.g., on the lagged values of these variables).

B. Measuring the Value of Conglomerates

Here, I discuss the various measures I use at both the firm level and the macro level to determine the value of conglomerates relative to the value of focused firms.

1. Measures at the Firm Level

My measure of the relative value of conglomerates at the firm level is conglomerate excess value (EV). I follow the Berger and Ofek's (1995) method and calculate EV as the natural logarithm of the ratio of a conglomerate's market value over its imputed value. (3) The conglomerate's imputed value is calculated as the sum of the imputed values of the conglomerate's segments. (4) In the diversification discount literature, both sales and assets multiples are used to calculate the imputed values of segments. Since the empirical results under sales and assets multiples are similar, I report in the article only the results based on sales multiples. I report the means and medians of EV in Table I grouped by years, and in Table II for the whole sample. Consistent with the diversification discount literature, I find that both the mean and median EVs of conglomerates are negative and smaller than the mean and median EVs of focused firms.

I also calculate separately EV for bank-dependent conglomerates and bank-independent conglomerates. Bank-dependency is used to measure the degree of financial constraint faced by each firm in external capital markets. I classify those firms as bank-independent if they have Standard & Poor's bond ratings according to Compustat, and those firms as bank-dependent if they have no S&P bond ratings. As suggested in the literature, bank-dependent firms lack access to the commercial paper and bond markets. Thus, they are more credit constrained in external capital markets and also are more affected by the tightening of external markets compared to bank-independent firms (see, e.g., Kashyap, Lamont, and Stein, 1994).

My calculation of EV for the subsamples of bank-dependent and bank-independent conglomerates is similar to the calculation of EV based on the full sample of conglomerates, except for the choice of benchmarks. For bank-dependent conglomerates, to calculate imputed values and further EVs, I use bank-dependent focused firms as the benchmark. For bank-independent conglomerates, I use bank-independent focused firms as the benchmark. By doing so, I ensure that EV captures only the diversification effect (i.e., the difference between diversification and focus) rather than the bank-dependency effect (i.e., the difference between conglomerates and focused firms in terms of their bank-dependency). Table II reports the sample statistics of EV for both subsamples of conglomerates. It shows that the average EV of bank-independent conglomerates is not significantly different from the average EV of bank-dependent conglomerates.

2. Measures at the Macro Level

The relative value of conglomerates at the macro level reflects the value difference between the organization structure of conglomerates and the structure of focused firms. In other words, it can be viewed as the value of diversification.

I use several metrics to measure the value of conglomerates at the macro level. I first construct CI, an index of conglomerate return, based on conglomerate excess value, EV. The EV of each conglomerate measures the value difference between a conglomerate and a conceptually imputed matching firm, where the matching firm can be created by investing in the median focused firm in the industries in which the conglomerate operates. Accordingly, I calculate CI as the value difference between a portfolio of all the conglomerates in my sample and a portfolio of the matching firms of these conglomerates. I calculate an equal-weighted index (EWCI) and a value-weighted index (VWCI), using the relevant portfolios on an equal-weighted or a value-weighted basis. For VWCI, I assign the same weight to both the conglomerate and the conglomerate's matching firm, where the weight is measured as the ratio of the value of the conglomerate to the total values of all conglomerates at the beginning of each year.

In particular, to calculate EWC[I.sub.t] and VW[C.sub.t] in each year t, I first construct a portfolio of conglomerates by selecting those firms that are classified as diversified firms and which have data available at both the beginning and end of the year. I rebalance the portfolio of conglomerates at the beginning of each year to add new conglomerates and to reset weights. The portfolio of matching firms consists of the matching firms of all conglomerates. I construct the matching firm of each conglomerate at the beginning of each year, and calculate its value as the sum of the imputed values of the conglomerate's segments. Next, I calculate either value-weighted or equal-weighted average returns (i.e., the changes in firm values) for both the portfolio of conglomerates and the portfolio of matching firms for an investment period from the benchmark year to year t. EWC[I.sub.t] and VWC[I.sub.t] are the logarithms of the ratios of the average returns of these two portfolios. I treat 1984 as the benchmark year, so that both EWC[I.sub.t = 1984] and WWC[I.sub.t = 1984] equal zero.

The following numerical example illustrates the algorithm for EWC[I.sub.t] and VWC[I.sub.t]. Suppose there are two conglomerates in the sample, conglomerates A and B, and two imputed matching firms, C and D, for conglomerates A and B, respectively. Each matching firm is constructed as a combination of the median focused firms related to the conglomerate's segments. The market values of conglomerates A and B are $100 and $200 at t=0, respectively, and $150 and $250 at t=1, respectively. The matching firms C and D are valued at $120 and $220 at t=0, respectively, and $160 and $280 at t=1, respectively. In this case, the equally weighted index of conglomerate return at t=1 is:

EWC[I.sub.t = 1] = log (150/100+250/200 / 160/120+180/220) = 0.054, (1)

if EWC[I.sub.t = 0] = 0. On the other hand, if VWC[I.sub.t = 0] = 0, the value-weighted index at t=1 is:

VWC[I.sub.t = 1] = log [(150+250)/(100+200) / 100/100+200 160/120 + 200/100+200 280/220] = 0.031. (2)

I also construct another index of conglomerate return RI to directly measure the difference in the returns between conglomerates and focused firms. I calculate R[I.sub.t] as the logarithm of the ratio of the equal-weighted returns from the benchmark year to year t between the portfolio of all conglomerates and the portfolio of all focused firms. (5) I rebalance both portfolios at the beginning of each year to reset the weights and to add new conglomerates and new focused firms. Again, I use 1984 as the benchmark year, so that R[I.sub.t = 1984] = 0. Compared to EWCI and VWCI, RI is a more direct measure of the difference between conglomerates and focused firms. However, RI could be affected not only by the difference between diversification and focus, but also by the different industry clustering between the conglomerate and the focused-firm portfolios. In comparison, EWCI and VWCI are industry adjusted and therefore not affected by any industry effect. Also, the calculations of EWCI and VWCI are consistent with the calculation of EV in the firm-level study. To ensure the robustness of my results, I use all three measures in my macro-level analysis.

The annual levels of EWCI, VWCI, and RI are reported in Table I, Panel B, and plotted in Figure 1. Table I also shows the aggregate level of diversification in each year, based on all conglomerates and focused firms in that year. I measure the aggregate level of diversification in a given year by using the average number of unrelated segments in a firm and the proportion of conglomerates in the sample.

[FIGURE 1 OMITTED]

Several observations stand out. First, EWCI, VWCI, and RI show similar patterns over time, although VWCI exhibits more volatility. Second, all three measures show that the value of conglomerates relative to their focused peers declines over my sample period, especially during the 1990s. For example, consider a yearly rebalanced equally weighted portfolio of conglomerates and a similarly constructed portfolio of matching firms (created from related focused firms). From 1984 to 1997, the return of the conglomerate portfolio is only 82.9% of the return of the portfolio of matching firms, indicating a significant underperformance for conglomerates. As suggested in Bhide (1990), since the 1980s, the increased sophistication and efficiency of external capital markets has largely eliminated the advantages of internal markets. Thus, the declining value of conglomerates over time could be attributed to the maturity of financial markets. Third, according to Table I, the aggregate level of diversification decreases steadily in the same period. The average number of unrelated segments in a firm decreases from 1.69 in 1984 to 1.24 in 1997. The correlations between the indexes of conglomerate return and the level of diversification are positive and significant (not shown in the table). These correlations suggest that the trend of re-focusing starting in the 1980s might have been driven by the deteriorated value of the conglomerate form of organization structure relative to the structure of focused firms.

I further disaggregate the sample of conglomerates and focused firms into two subsamples, a bank-dependent subsample and a bank-independent subsample. I calculate EWCI, VWCI, and RI for conglomerates in both subsamples. Similar to the corresponding firm-level measures, bank-dependent focused firms and bank-independent focused firms are chosen as the benchmarks for bank-dependent and bank-independent conglomerates, respectively. Table I, Panel B, reports the annual levels of EWCI for conglomerates in both subsamples, and Figure 1, Panel B, plots their trends.

Figure 1 shows that bank-dependent and bank-independent conglomerates exhibit different trends over time. The EWCI of bank-dependent conglomerates decreases steadily from 1984 to 1997. This trend is similar to the EWCI of the whole sample of conglomerates. In contrast, the EWCI of bank-independent conglomerates increases in the late 1980s and, unlike the EWCI of bank-dependent conglomerates, does not show a significant downward trend in the 1990s. I also find similar results based on the VWCI and RI of conglomerates (not reported here). These results are consistent with the recent findings in Akbulut and Matsusaka (2005). Based on the announcement effect of diversifying mergers, Akbulut and Matsusaka (2005) find that diversification gradually fell out of favor and that small firms fell more than large firms. In general, small firms are more bank-dependent than large firms. (6) Thus, my results on the different trends between bank-dependent and bank-independent conglomerates are consistent with the results in Akbulut and Matsusaka (2005).

One explanation for such different trends is the different sensitivities of bank-dependent and bank-independent conglomerates as they respond to the development of external capital markets. In general, bank-dependent conglomerates face costly external financing constraints, so they rely more on internal capital markets as a cheap financing alternative than on external capital markets. When external capital markets become more efficient, internal capital markets gradually lose their cost advantage. As a result, for bank-dependent conglomerates, the value of diversification gradually declines.

On the other hand, bank-independent conglomerates have easy access to external capital, so that internal capital markets are less critical for this type of conglomerate than for the bank-dependent conglomerate. Consequently, the value of bank-independent conglomerates is less sensitive to both the increased efficiency of external markets and the reduced advantage of internal markets.

C. Construction of Other Variables

I measure capital market conditions by using the following macro variables. I use the Federal Funds rate (FEDRT), the discount rate charged by the Federal Reserve (DSCRT), and the logarithm of the M2 money supply (M2) to measure the tightness of monetary policy. I use the price-to-earnings ratio of the S&P 500 (SPPE), which I obtain from Shiller (2000), to capture financing conditions in equity market. To capture credit conditions in credit markets, I use the interest rate of BAA rated bonds (BAA) and the interest rate of six-month commercial paper (CPRT). All the interest rates are annual rates averaged based on monthly data. To measure the general activities of external financing, I use the logarithm of gross equity issuance (EQUITY) and the logarithm of gross long-term debt issuance (DEBT) in each year, obtained from the Federal Reserve Bulletin. I use these macro variables to measure the cost of financing at the aggregate level in external capital markets. The lower the SPPE, the more likely it is that firms' equity is underpriced and thus the more expensive is equity financing. Similarly, a higher interest rate, i.e., a higher BAA or CPRT, indicates more expensive debt financing; a tighter monetary policy, i.e., a lower M2 or higher FEDRT or DSCRT, indicates a reduced general availability of capital in capital markets; and a lower level of external financing activities, i.e., lower EQUITY or DEBT, indicates more depressed markets for external financing.

I also construct the following control variables for the regressions in firm-level studies. I calculate firm size (SIZE) as the logarithm of the market value of total assets, where the market value of total assets is the market value of equity plus the book value of liabilities. I calculate the long-term debt ratio (LTDR) as the ratio of the amount of long-term debt to the book value of total assets. CAPEX is capital expenditures scaled by sales revenue, and OPINC is operating income before interest and taxes, divided by sales revenue. Table II presents the sample statistics of the control variables and the macro variables. It shows that conglomerates are less bank-dependent and larger in size than focused firms.

II. Value of Conglomerates and Capital Market Conditions

In this section, I study the relation between the value of conglomerates and capital market conditions at both the macro and the firm levels.

A. Motivational Evidence

Here, I study the pair-wise correlations between the value of conglomerate and capital market conditions at the macro level. Three indexes of conglomerate return, EWCI, VWCI, and RI, are used to measure the relative value of conglomerates at the macro level. Table III presents the correlation between these three indexes and the macro variables measuring capital market conditions, as well as the correlations among the macro variables.

Table III shows that all three indexes are positively correlated with the cost of financing in external capital markets. EWCI, VWCI, and RI are all negatively correlated with SPPE, M2, EQUITY, and DEBT, and positively correlated with BAA, CPRT, FEDRT, and DSCRT. All the correlations are significant at the meaningful levels. These correlations indicate that the value of diversification increases when external capital markets are more constrained and external financing is more costly at the aggregate level.

I further disaggregate my sample to bank-dependent and bank-independent subsamples. Bank-dependency is used as my proxy for the degree of financial constraint faced by each conglomerate in external capital markets. I calculate the correlations between the macro variables and EWCI for bank-dependent and bank-independent conglomerates separately. The results are presented in Table IV, Panel A. I also calculate similar correlations based on VWCI and RI, and present these correlations in Panels B and C of Table IV.

I find that both EWCI and RI of bank-dependent conglomerates are negatively correlated with SPPE, M2, EQUITY, and DEBT, and positively correlated with BAA, CPRT, FEDRT, and DSCRT. All the correlations are significant. In comparison, both the EWCI and RI of bankindependent conglomerates are uncorrelated with the macro variables. Further, both the VWCI of bank-dependent conglomerates and the VWCI of bank-independent conglomerates are negatively correlated with the tightness of external capital markets. However, the correlations for bank-dependent conglomerates are more significant, both economically and statistically, than are those for bank-independent conglomerates.

In general, my results show that restrictive market-level financing conditions increase the value of bank-dependent conglomerates relative to their focused peers, but they have a less significant impact on the value of bank-independent conglomerates. Thus, my results are consistent with my second hypothesis, and support the notion that the relative value of financially constrained conglomerates is more sensitive to external financing conditions than are financially unconstrained conglomerates.

B. Empirical Evidence at the Firm Level

In this section, I study the relation between the value of conglomerates and capital market conditions at the firm level. I test the first hypothesis based on conglomerate excess value E[V.sub.i,t], using the full sample of conglomerates. I run the following regression:

E[V.subi,t] = [alpha] + [[beta].sub.1]MK[T.sub.t] + [[beta].sub.2][x.sub.i, t-1] + [[epsilon].sub.i, t] (3)

where MK[T.sub.t] stands for a macro variable selected from SPPE, M2, EQUITY, DEBT, BAA, CPRT, FEDRT, and DSCRT. [x.sub.i.t.l] is a vector of control variables consisting of firm size ([SIZE.sub.i.t.l]), operating income (OPIN[C.sub.i,t-l]), capital expenditures (CAPE[X.sub.i,t-l]), and long-term debt ratio (LTDR.sub.i.t.l]). (7) All control variables are lagged by one year to avoid the collinearity between the control variables and MK[T.sub.t]. For Regression (3) and also for the other regressions later in the article, I allow correlated residuals within each cross section. I conduct significance tests by using the Huber-White heteroskedasticity-consistent standard errors. In all regressions, time effect is not controlled, since the macro variables vary only in the time dimension.

In Regression (3), [[beta].sub.1] measures the sensitivity of conglomerate excess value to financing conditions in external capital markets. According to the first hypothesis, I expect [[beta].sub.1] to be negative for the macro variables SPPE, M2, EQUITY, and DEBT, and positive for the macro variables BAA, CPRT, FEDRT, and DSCRT. Table V presents the results from this regression. It shows that [[beta].sub.1] follows the expected signs for all the macro variables and is significant for all macro variables except DSCRT. Also, the impact of capital market conditions on conglomerate excess value is economically significant. A one-standard-deviation change in a macro variable would cause the excess value of an average conglomerate to change by at least 14%. For example, a one-standard-deviation increase in SPPE would result in a decrease in EV by 0.034, which is 61% of the average EV (-0.057). Thus, my regression results on EV support the first hypothesis. The results suggest that when external financing becomes more costly at the aggregate level, the value of a conglomerate increases relative to its focused peers.

In testing the second hypothesis, I use bank-dependency (DEP) as my proxy for the degree of financial constraint faced by each conglomerate. I pool the EV constructed in the bank-dependent subsample with EV constructed in the bank-independent subsample, and run the following regression:

E[V.sub.i, t] = [alpha] + [[gamma].sub.1]MK[T.sub.t] * (1 - DEP) + [[gamma].sub.2]MK[T.sub.t] * DEP + [[gamma].sub.2][x.sub.i, t-1] + [[gamma].sub.4] DEP + [[epsilon].sub.i, t]. (4)

The construction of the control variables [x.sub.i,t-l] and the macro variable MK[T.sub.t] are the same as those discussed earlier. Here, 71, the coefficient of the interaction term between MK[T.sub.t] and the dummy of bank-independency 1-DEP, measures the sensitivity of excess value to capital market conditions for bank-independent (i.e., financially unconstrained) conglomerates. [[gamma].sub.2], the coefficient of the interaction term between MK[T.sub.t] and the dummy of bank-dependency DEP, measures the sensitivity for bank-dependent (i.e., financially constrained) conglomerates. According to my second hypothesis, I expect [[gamma].sub.2] to be negative and smaller than [[gamma].sub.1] for macro variables SPPE, M2, EQUITY, and DEBT, and positive and larger than [[gamma]sub.1] for macro variables BAA, CPRT, FEDRT, and DSCRT.

Table VI provides the results from Regression (4). For all the macro variables, [[gamma].sub.2] is significant and has the signs predicted by the second hypothesis. In contrast, [[gamma].sub.1] is only significant and negative for SPPE and M2, and is insignificant for the other macro variables. For example, an increase of BAA by one standard deviation would cause the EV of an average bank-dependent conglomerate to increase by 0.045 (78% of average EV), but an increase of BAA would not change the EV of an average bank-independent conglomerate.

I also test whether the sensitivities to market conditions are different between bank-dependent and bank-independent conglomerates, i.e., [[gamma].sub.2] - [[gamma].sub.1] = 0. My tests show that [[gamma].sub.2] - [[gamma].sub.1] is significantly different from zero for EQUITY, DEBT, CPRT, FEDRT, and DSCRT. However, [[gamma].sub.2] - [[gamma].sub.2] is insignificant for SPPE, M2, and BAA. Thus, my results are generally consistent with the second hypothesis, suggesting that the value of a conglomerate increases more in tightened capital markets if the conglomerate is bank-dependent and thus more financially constrained in external capital markets.

Overall, my evidence at both the macro level and the firm level suggests that when the cost of external funds is high or external capital is scarce, diversification is more valuable relative to separation, and particularly so for financially constrained conglomerates such as bank-dependent conglomerates. (8)

III. Robustness Checks

In the following section, I perform several additional tests to check the robustness of my results.

A. Alternative Specifications with Differenced Variables

My first robustness check tests the possibility of spurious regressions due to trended and nonstationary time series. According to Figure 1, the value of conglomerates exhibits strong downward time trends. The macro variables also display either downward or upward trends over time. When I perform Dickey-Fuller tests for unit roots of my time-series variables, the results show that my conglomerate index measures, EWCI and VWCI, are both integrated of order one, i.e., I(1), and RI is integrated of order two, i.e., I(2). The macro variables SPPE, M2, EQUITY, CPRT, FEDRT, and DSCRT are also nonstationary and integrated of order one or of higher orders, while the macro variables DEBT and BAA are stationary.

My nonstationary conglomerate value and macro variables may lead to spurious regressions if both variables have different embedded trends (regardless of a stochastic or deterministic trend). If this is the case, then the estimated relation between conglomerate value and macro variables could be a consequence of the underlying trends in these variables, rather than a true relation. In this case, to draw an appropriate inference, I need to transform the nonstationary conglomerate value variables and the nonstationary macro variables into stationary ones. To do so, I calculate the first difference for both E[V.sub.i,t] and MK[T.sub.t] to create [DELTA]E[V.sub.i,t] and [DELTA]MK[T.sub.t], respectively; [DELTA]E[V.sub.i,t] = E[V.sub.i,t] - E[V.sub.i,t-1] and [DELTA]MK[T.sub.t] = MK[T.sub.i,t-1] - MK[T.sub.i,t-1]. (9) Then, I run the following regression to test the first hypothesis:

[DELTA]E[V.sub.i,t] = [alpha] + [[beta].sub.1] [DELTA]MK[T.sub.t] + [[beta].sub.2] [x.sub.i,t-1] + [[epsilon].sub.i,t] (5)

In this regression, I control for firm fixed effects, considering that each firm follows a firm-specific trend. The results from this new regression are presented in Table VII. I find that [[beta].sub.1], the coefficient of [DELTA]MK[T.sub.t], is negative and significant for the macro variables [DELTA]SPPE, [DELTA]M2, [DELTA]EQUITY, and [DELTA]DEBT, and positive and significant for [DELTA]BAA and [DELTA]CPRT. However, [[beta].sub.1] is insignificant for [DELTA]FEDRT and [DELTA]DSCRT, although it has the expected positive sign. Thus, my results based on differenced excess value [DELTA]E[V.sub.i,t], although slightly weaker, are similar to those based on E[V.sub.i,t]. They are generally consistent with the first hypothesis.

I also check the robustness of my results for the second hypothesis by running a regression similar to Regression (4), but using differenced variables [DELTA]E[V.sub.i,t] and [DELTA]MK[T.sub.t]:

[DELTA]E[V.sub.i,t] = [alpha] + [[gamma].sub.1] [DELTA]MK[T.sub.t] * (1 - DEP) + [[gamma].sub.2] [DELTA]MK[T.sub.t] * DEP + [[gamma].sub.3][x.sub.i,t-1] + [[epsilon].sub.i,t]. (6)

Table VIII presents these results. According to Table VIII, [[gamma].sub.2] is negative and significant for macro variables [DELTA]SPPE, [DELTA]M2, [DELTA]EQUITY, and [DELTA]DEBT, and positive and significant for [DELTA]BAA, [DELTA]CPRT, [DELTA]FEDRT, and [DELTA]DSCRT. The results on [[gamma].sub.2] suggest that if there is a negative market shock to external financing, the value of a bank-dependent conglomerate increases relative to its focused peers. Further, my tests on [[gamma].sub.2] - [[gamma].sub.1] show that [[gamma].sub.2] is significantly different from [[gamma].sub.1] for all macro variables except [DELTA]SPPE. Thus, these results are consistent with the second hypothesis, and suggest that the values of financially constrained conglomerates are more sensitive to shocks in external capital markets than are conglomerates that are not financially constrained.

It is worth noting that the above regressions based on differenced variables (which I use to transform nonstationary processes to stationary processes) could be inefficient, in the sense that they may exclude part of the dynamics between conglomerate value and capital market conditions. As discussed earlier, estimating regressions with nonstationary variables may produce a spurious relation. However, there is another possibility. If both variables share a common trend rather than different trends or, in other words, if they are cointegrated, then estimating regressions with nonstationary variables could capture the long-run relation between these variables. This long-run relation is in fact predicted by current theories on internal capital markets: i.e., as external markets become less costly over the long run, the internal capital markets featured in conglomerates become less valuable.

The regressions based on differenced variables would miss the long-run relation between conglomerate value and market conditions if these two variables are cointegrated. Instead, such regressions could capture only the short-term dynamics, i.e., the relation between the deviation of a firm's conglomerate value from its long-run trend and the deviation of a macro condition from its long-run trend. Nevertheless, my results in this section confirm the robustness of my earlier findings. They show that the condition of capital markets has a significant impact on conglomerate value, even if I ignore the potential long-run relation between them.

My results from Regressions (5) and (6) might also address the concern of selection bias. Many studies on the diversification discount challenge the notion that conglomerate excess value measures the value difference between conglomerates and focused firms. Their concern is that conglomerate excess value could be subject to selection bias, i.e., the pre-diversification poor performance of diversifying firms and/or target firms. Regressions that use [DELTA]E[V.sub.i,t] as the dependent variable could alleviate this concern. [DELTA]E[V.sub.i,t] can be viewed as measuring the difference between a conglomerate's return and its imputed return in year t, where the conglomerate's return in year t is the change in the value of the conglomerate from year t-1 to year t, and the conglomerate's imputed return is the change in its imputed value from year t-1 to year t. Therefore, [DELTA]E[V.sub.i,t] is less prone to the selection bias than is E[V.sub.i,t], since [DELTA]E[V.sub.i,t] captures the difference between conglomerates and focused firms during year t and is less affected by the difference between them prior to year t such as the pre-diversification difference. (10)

B. Alternative Measures of Financial Constraints

In this robustness check, I test the second hypothesis by using two other measures of financial constraint. First, I use a conglomerate's firm size, SIZ[E.sub.t-1], as an alternative proxy for the extent of the financial constraint faced by the conglomerate. In doing so I follow Gertler and Gilchrist (1994), who find that sales, inventory, and debt growth rates of small manufacturers are more sensitive to recessions than are those of large manufacturers. Thus, small conglomerates are clearly more financially constrained in tightened external capital markets than are large conglomerates.

The new regression is similar to that in Regression (4), except that I replace the variable 1-DEP with the dummy for large firm size and the variable DEP with the dummy for small firm size. The dummy for small size equals one if a conglomerate's lagged firm size SIZ[E.sub.t-1] is below the median level of SIZ[E.sub.t-1] of all conglomerates in my sample, and the dummy for large size equals one if a conglomerate's SIZ[E.sub.t-1] is above the median SIZ[E.sub.t-1]. In these regressions, [[gamma].sub.2] measures the sensitivity of conglomerate excess value to capital market conditions for small conglomerates (financially constrained) and [[gamma].sub.1] measures the sensitivity for large conglomerates (financially unconstrained). According to the second hypothesis, I expect [[gamma].sub.2] and [[gamma].sub.2] - [[gamma].sub.1] to be negative for the macro variables SPPE, M2, EQUITY, and DEBT, and positive for the macro variables BAA, CPRT, FEDRT, and DSCRT.

Table IX presents the results from these new regressions. I find that [[gamma].sub.2] is significant and follows the expected signs for all the macro variables. Thus, for a small conglomerate (with small SIZ[E.sub.t-1]), its excess value increases when external funds become more costly. In contrast, [[gamma].sub.1] is insignificant for most macro variables, suggesting that the excess value of a large conglomerate (with large SIZ[E.sub.t-1]) is not sensitive to the change in capital market conditions. My tests on [[gamma].sub.2] - [[gamma].sub.1] also show that [[gamma].sub.2] is significantly different from [[gamma].sub.1] for all the macro variables. Again, these results using SIZ[E.sub.t-1] as the measure of financial constraint are consistent with the second hypothesis and confirm the robustness of my earlier results. (11)

I also use the Kaplan and Zingales (1997) index to measure the extent of the financial constraint faced by a firm. I construct a KZ index at time t using the coefficients in Lamont, Polk, and Saa-Requejo (2001). (12)

K[Z.sub.t] = -1.002 Cash flow / [K.sub.t-1] + 0.283 M[B.sub.t-1] + 3.139 Debt / Total capital -39.368 Dividends / [K.sub.t-1] -1.315 Cash / [K.sub.t-1]. (7)

A higher level of K[Z.sub.t] indicates that a firm is more financially constrained. In the new regressions based on K[Z.sub.t], I use specifications similar to that in Regression (4), but I replace the variable 1-DEP with the low KZ dummy and the variable DEP with the high KZ dummy. The low KZ dummy equals one if a conglomerate's K[Z.sub.t] is below the median level of K[Z.sub.t] of all conglomerates in my sample, and the high KZ dummy equals one if a conglomerate's K[Z.sub.t] is above the median K[Z.sub.t]. Table X presents the results from this robustness check.

In general, for conglomerates with high KZ, the excess values increase when there is a negative shock to external financing, but the conglomerates with low KZ are less sensitive to such shocks. Again, my results based on K[Z.sub.t] as the measure of financial constraint support the second hypothesis and confirm the robustness of my earlier results.

IV. Conclusion

In this article, I study the variation in the value of conglomerates over time and under various capital market conditions.

I first construct various annual indexes of conglomerate return to measure the value of conglomerates relative to focused firms at the aggregate level. Based on the conglomerate return indexes, I find that over the period 1984-1997, the value of conglomerates declines relative to focused firms. I also find that the relative value of bank-dependent conglomerates declines at a faster rate than does the relative value of bank-independent conglomerates.

Next, I study how the overall capital market conditions affect the value of conglomerates relative to their focused peers (i.e., the value of diversification), and find a significant impact. In general, the relative value of a conglomerate increases when external funds become more costly. Also, such an increase in the value of diversification is greater for financially constrained conglomerates, e.g., for bank-dependent conglomerates or small conglomerates, than for conglomerates that are not financially constrained. Thus, my results confirm the theories on the advantage of diversification over focus, i.e., the ability of conglomerates to substitute costly external capital markets with internal capital markets. My findings suggest that when the cost of external funding is high, the availability of internal capital market is value-enhancing, especially for those conglomerates that are financially constrained.

Although the primary purpose of my article is to study the variation in the value of diversification, my evidence also illuminates the trends of diversifying mergers and focus-increasing divestitures over time. Many studies document a decline in diversifying activities beginning around 1980s. My findings on the value of diversification and capital market conditions suggest that market conditions for external financing could play a role in this trend. Over time, capital markets have grown more accessible and more developed due to deregulation, competitions, and various other factors. The increased efficiency of external capital markets has reduced the importance of internal capital markets and thus the value of diversification. The reduced value of diversification might have contributed to the recent decline in diversifications and increase in refocusing.

References

Akbulut, M.E. and J.G. Matsusaka, 2005, "The Waning of Corporate Diversifications," University of Southern California Working Paper.

Berger, P.G. and E. Ofek, 1995, "Diversification's Effect on Firm Value," Journal of Financial Economics 37, 39-65.

Bhide, A., 1990, "Reversing Corporate Diversification," Journal of Applied Corporate Finance 3, 70-81.

Campa, J.M. and S. Kedia, 2002, "Explaining the Diversification Discount," Journal of Finance 57, 1731-1762.

Chevalier, J., 2004, "What Do We Know About Cross-subsidization? Evidence from Merging Finns," Advances in Economic Analysis and Policy 4, Article 3.

Doukas, J.A. and B.K. Ozgur, 2004, "Excess Cash Flows and Diversification Discount," Financial Management 33, 71-88.

Gertler, M. and S. Gilchrist, 1994, "Monetary Policy, Business Cycles, and the Behavior of Small Manufacturing Finns," Quarterly Journal of Economics 109, 309-340.

Gertner, R.H., D.S. Scharfstein, and J. Stein, 1994, "Internal versus External Capital Markets," Quarterly Journal of Economics 109, 1211-1230.

Graham, J.R., M.L. Lemmon, and J. Wolf, 2002, "Does Corporate Diversification Destroy Value?" Journal of Finance 57, 695-720.

Hubbard, R.G. and D. Palia, 1999, "A Reexamination of the Conglomerate Merger Wave in the 1960s: An Internal Capital Markets View," Journal of Finance 54, 1131-1152.

Jandik T. and A.K. Makhija, 2005, "Can Diversification Create Value? Evidence from the Electric Utility Industry," Financial Management 34, 61-93.

Kaplan, S.N. and L. Zingales, 1997, "Do Investment-Cash Flow Sensitivities Provide Useful Measures of Financing Constraints?" Quarterly Journal of Economics 112, 169-215.

Kashyap, A.K., O.A. Lamont, and J.C. Stein, 1994, "Credit Conditions and the Cyclical Behavior of Inventories," Quarterly Journal of Economics 109, 565-592.

Lamont, O., 1997, "Cash Flow and Investment: Evidence from Internal Capital Markets," Journal of Finance 52, 83-109.

Lamont, O., C. Polk, and J. Saa-Requejo, 2001, "Financial Constraints and Stock Returns," Review of Financial Studies 14, 529-554.

Lang, L. and R.M. Stulz, 1994, "Tobin's q, Corporate Diversification, and Firm Performance," Journal of Political Economics 102, 1248-1280.

Lins, K.V. and H. Servaes, 2002, "Is Corporate Diversification Beneficial in Emerging Market?" Financial Management 31, 5-31.

Matsusaka, J.G., 1993. "Takeover Motives during the Conglomerate Merger Wave," RAND Journal of Economics 24, 357-379.

Matsusaka, J.G. and V. Nanda, 2002, "Internal Capital Markets and Corporate Refocusing," Journal of Financial Intermediation 11,176-211.

Rajan, R.G., H. Servaes, and L. Zingales, 2000, "The Cost of Diversity: The Diversification Discount and Inefficient Investment," Journal of Finance 60, 35-80.

Scharfstein, D.S. and J. Stein, 2000, "The Dark Side of Internal Capital Markets: Divisional Rent Seeking and Inefficient Investments," Journal of Finance 55, 2537-2564.

Servaes, H., 1996, "The Value of Diversification during the Conglomerate Merge Wave," Journal of Finance 51, 1201-1225.

Shiller, R., 2000, Irrational Exuberance, Princeton, NJ, Princeton University Press and Broadway Books.

Shin, H. and R.M. Stulz, 1998, "Are Internal Capital Markets Efficient?" Quarterly Journal of Economics 113, 531-552.

Stein, J.C., 1997, "Internal Capital Markets and the Competition for Corporate Resources," Journal of Finance 52, 111-133.

Villalonga, B., 1999, "Does Diversification Cause the Diversification Discount?" Financial Management 33, 5-27.

Villalonga, B., 2004, "Diversification Discount or Premium? New Evidence from BITS Establishment-Level Data," Journal of Finance 59, 479-506.

Whited, T.M., 2001, "Is It Inefficient Investment that Causes the Diversification Discount?" Journal of Finance 56, 1667-1691.

Williamson, O.E., 1975, Markets and Hierarchies: Analysis and Antitrust Implications, New York, NY, Free Press.

(1) There is also a large theoretical literature on internal capital markets. Under the efficiency view, Stein (1997) argues that headquarters on the top of the conglomerate hierarchy can use their information advantage and allocate funds more efficiently through winner-picking in internal capital markets compared to capital allocation in external markets (see also Williamson, 1975, and Gertner, Scharfstein, and Stein, 1994). However, the inefficiency view argues that internal capital markets open opportunities for distorted investment due to the politics within conglomerates (Rajah, Servaes, and Zingales, 2000, and Scharfstein and Stein, 2000). Matsusaka and Nanda (2002) suggest that the relative value of diversification versus separation depends on the trade off between the cost of raising external funds and the inefficiency of internal capital markets.

(2) In a robustness check, I define focused and diversified firms based on the number of reported segments. The results in the article remain mostly unchanged under this alternative definition.

(3) Some recent studies suggest that excess value could be a biased estimate to measure conglomerate value (see, e.g., Graham, Lemmon, and Wolf, 2002, Whited, 2001, Campa and Kedia, 2002, and Villalonga, 1999, 2004). I will address this concern later in Section III.A.

(4) I calculate the imputed value of segments as the product of segment sales and the value-to-sales ratio of the median focused firm in the segment's industry. Industry matching is performed based on the narrowest of a four, three, or two-digit SIC code grouping that has at least five focused firms with sales greater than or equal to $20 million and with non-missing values for the required variables. I winsorize excess values at the 1% level in both tails of the distribution.

(5) As a robustness check, I also construct RI using value-weighted portfolios. The results based on value-weighted RI are similar to those based on the equal-weighted RI.

(6) In my sample, 68.3% of small conglomerates (with capitalizations in the bottom three quartiles of the sample) are bank-dependent, compared to 27.3% of large conglomerates (with capitalizations in the top quartile). This sample statistic is not included in the article, but is available upon request.

(7)I control for SIZE, OPINC, and CAPEX to be consistent with Berger and Ofek (1995), and I control for LTDR since previous literature suggests that firms with different leverages respond differently to market shocks.

(8) Note that some of my macro variables could also proxy the general business activities of product markets. Thus, some of my results on the impact of tightened capital markets could also be interpreted based on depressed product markets.

(9) By first differencing E[V.sub.i,t] and MK[T.sub.t], I assume that both variables follow stochastic trends, e.g., E[V.sub.i,t] = [[alpha].sub.i] + E[V.sub.i,t-1] + [u.sub.i], where [u.sub.t] is I(0) and a function of macro variables and control variables. However, if both variables follow deterministic trends, e.g., E[V.sub.i,t] = [[alpha].sub.i] + [[delta].sub.i]t + [u.sub.i], then first differencing could be inappropriate and instead detrending (i.e., computing the residuals from a regression on time) could be a right approach. I do not use detrending in the paper since different firms follow different trends and detrending at a firm level for each firm would be difficult to be implemented.

(10) However, it is possible that the pre-diversification difference between conglomerates and focused firms could affect or be related to their different behaviors during year t. In this sense, [DELTA]E[V.sub.i,t] may not be a perfect solution for the selection bias problem.

(11) I also compare large conglomerates with small conglomerates at the macro level, similar to the study in Section II.A based on bank-dependency. I find supporting evidence for the second hypothesis as well. However, because the macro-level study based on firm size could suffer from the benchmark problem, I choose not to present these results in the paper. Unlike the case with bank-dependency, where I can use a sample of bank-dependent focused firms as the benchmark for bank-dependent conglomerates, it is hard to construct a benchmark of focused firms with the same firm size as the size of each conglomerate. As a result, the conglomerate indexes for large and small conglomerates could capture both a diversification effect and a size effect. For the same reason, I do not present the results from the macro-level study on KZ index that I discuss below.

(12) Following the literature, cash flow is defined as (Compustat item 18 + item 14), K is capital expenditures defined as item 8, MB is market-to-book ratio equal to (item [25.sup.*]item 199 + item 6--item 60)/item 6, and cash is defined as item 1.

An Yan *

I thank Sris Chatterjee, Thomas Chemmanur, Gayane Hovakimian, and Murali Jagannathan, as well as seminar participants at Fordham University for helpful comments or discussions. I also give special thanks to Bill Christie (the Editor), and an anonymous referee for helpful suggestions. I am responsible for any errors or omissions.

* An Yan is an assistant professor at Fordham University in New York, NY.

Table I. Sample Distribution and Indexes of Conglomerate Return (Grouped by Years) The sample covers from 1984 to 1997. 1 define conglomerates as those firms that report at least two segments operating in different industries. I group segments in a firm using 4-digit SIC codes. I calculate excess value (EV) as the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments, and segment imputed values as the product of segment sales and the value to sales ratio of the median focused firm in the segment's industry. EWCI, VWCI, and RI are the indexes of conglomerate return. I calculate EWCI (VWCI) as the logarithm of the ratio of the equally weighted (value-weighted) average value changes between conglomerates and their matching firms; the value of the matching firm of each conglomerate as the sum of the imputed values of the conglomerate's segments as stand-alone fines; and RI as the logarithm of the ratio of the equally weighted value changes between a portfolio of conglomerates and a portfolio of focused firms. I define bank-independent firms as those firms with S&P bond ratings according to Compustat, and bank-dependent firms as those firms with no S&P bond ratings. Panel A. Sample Distribution and Annual Levels of Diversification based on the Whole Sample No. of No. of Focused No. of Year Firms Firms Conglomerates 1984 2,499 1,567 932 1985 2,461 1,610 851 1986 2,497 1,711 786 1987 2,665 1,910 755 1988 2,577 1,896 681 1989 2,524 1,886 638 1990 2,538 1,921 617 1991 2,658 2,031 627 1992 2,910 2,256 654 1993 3,257 2,605 652 1994 3,547 2,881 666 1995 3,893 3,209 684 1996 4,287 3,597 690 1997 4,356 3,716 640 Panel B. Conglomerate Indexes based on the Whole Sample or Subsamples Median EV of Year Conglomerates EWCI VWCI 1984 -0.117 0.000 0.000 1985 -0.142 -0.032 -0.004 1986 -0.108 -0.024 -0.006 1987 -0.027 0.011 0.058 1988 -0.076 -0.027 -0.027 1989 -0.060 -0.020 0.044 1990 -0.039 -0.022 0.082 1991 -0.062 -0.063 -0.002 1992 -0.049 -0.061 -0.021 1993 -0.036 -0.066 -0.082 1994 -0.066 -0.096 -0.105 1995 -0.042 -0.129 -0.163 1996 -0.048 -0.150 -0.142 1997 -0.058 -0.187 -0.155 Panel A. Sample Distribution and Annual Levels of Diversification based on the Whole Sample Percentage of Avg. No. of Mean EV of Year Conglomerates Segments Conglomerates 1984 0.373 1.687 -0.104 1985 0.346 1.624 -0.128 1986 0.315 1.548 -0.082 1987 0.283 1.483 -0.027 1988 0.264 1.440 -0.048 1989 0.253 1.419 -0.049 1990 0.243 1.402 -0.045 1991 0.236 1.386 -0.049 1992 0.225 1.370 -0.026 1993 0.200 1.330 -0.022 1994 0.188 1.312 -0.043 1995 0.176 1.298 -0.057 1996 0.161 1.264 -0.059 1997 0.147 1.238 -0.087 Panel B. Conglomerate Indexes based on the Whole Sample or Subsamples EWCI (bank- EWCI (bank- Year RI dependent) independent) 1984 0.000 0.000 0.000 1985 -0.004 -0.051 0.047 1986 -0.009 -0.084 0.062 1987 0.035 -0.017 0.065 1988 0.040 -0.056 0.106 1989 0.043 -0.075 0.130 1990 0.021 -0.064 0.041 1991 -0.067 -0.138 0.056 1992 -0.069 -0.149 0.113 1993 -0.066 -0.164 0.084 1994 -0.113 -0.200 0.059 1995 -0.165 -0.230 -0.034 1996 -0.184 -0.282 -0.037 1997 -0.160 -0.350 -0.018 Table II. Sample Statistics I define conglomerates as those firms that report at least two segments operating in different industries. I group segments in a firm using 4-digit SIC codes. I calculate excess value (EV) as the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. SIZE is the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC is operating income before interest and taxes deflated by sales revenue; CAPEX is capital expenditures deflated by sales revenue; and LTDR is the amount of long-term debt over the book value of assets. I define as bank-dependent those firms with no S&P bond ratings, and as bank-independent those firms with S&P bond ratings. Bank-dependency equals one for bank-dependent firms and zero for bank-independent firms. FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated corporate bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues in each year, respectively. Panel A. Based on a Full Sample Consisting of 32,796 Firm-Years of Focused Firms and 9,873 Firm-Years of Conglomerates Focused Firms Mean Median Std. Dev. EV 0.024 0.000 0.550 EV (Bank-dependent) 0.028 0.000 0.565 EV (Bank-independent) 0.023 0.000 0.487 Bank-dependency 0.759 1.000 0.428 SIZE 4.900 4.685 1.534 LTDR 0.180 0.131 0.181 CAPEX 0.079 0.042 0.120 OPINC 0.073 0.069 0.108 Conglomerates Mean Median Std. Dev. EV -0.057 -0.066 0.483 EV (Bank-dependent) -0.0991 -0.132 0.501 EV (Bank-independent) -0.105 -0.111 0.453 Bank-dependency 0.544 1.000 0.498 SIZE 5.806 5.685 1.854 LTDR 0.213 0.191 0.164 CAPEX 0.065 0.041 0.086 OPINC 0.069 0.068 0.079 Panel B. Based on a Sample of Annual Levels of Macro Variables from 1984 to 1997 Mean Median Std. Dev. SPPE 18.246 17.267 5.804 M2 8.045 8.098 0.168 EQUITY 5.178 5.263 0.446 DEBT 7.006 7.074 0.371 BAA 9.905 8.977 1.860 CPRT 6.476 6.163 1.905 FEDRT 6.407 6.247 2.101 DSCRT 5.650 5.555 1.671 Table III. Correlations Among Indexes of Conglomerate Returns and Macro Variables I define conglomerates as those firms that report at least two segments operating in different industries. I group segments in a firm using 4-digit SIC codes. EWCI, VWCI, and RI are the indexes of conglomerate return. I calculate EWCI (VWCI) as the logarithm of the ratio of the equal-weighted (value-weighted) average value changes between conglomerates and their matching firms; the value of the matching firm of each conglomerate as the sum of the imputed values of the conglomerate's segments as stand-alone firms; and RI as the logarithm of the ratio of the equal-weighted value changes between a portfolio of conglomerates and a portfolio of focused firms. FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated corporate bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues in each year, respectively. EWCI VWCI RI SPPE SPPE -0.909 *** -0.740 *** -0.818 *** 1 [0.000] [0.003] [0.000] M2 -0.799 *** -0.611 ** -0.724 *** 0.924 *** [0.001] [0.020] [0.003] [0.000] EQUITY -0.624 ** -0.652 ** -0.648 ** 0.660 *** [0.02] [0.011] [0.012] [0.010] DEBT -0.615 ** -0.544 ** -0.524 * 0.791 *** [0.019] [0.044] [0.054] [0.001] BAA 0.746 *** 0.635 ** 0.703 *** -0.887 *** [0.002] [0.015] [0.005] [0.000] CPRT 0.546 ** 0.537 ** 0.615 ** -0.639 ** [0.044] [0.048] [0.019] [0.014] FEDRT 0.552 ** 0.555 ** 0.627 ** -0.639 * [0.041] [0.040] [0.017] [0.014] DSCRT 0.511 * 0.512 * 0.558 ** -0.656 ** [0.062] [0.061] [0.038] [0.011] M2 EQUITY DEBT BAA SPPE M2 1 EQUITY 0.545 ** 1 [0.044] DEBT 0.821 *** 0.769 *** 1 [0.000] [0.001] BAA -0.571 ** -0.696 *** -0.928 *** 1 [0.033] [0.006] [0.000] CPRT -0.356 -0.808 *** -0.819 *** 0.827 *** [0.212] [0.000] [0.000] [0.000] FEDRT -0.715 *** -0.787 *** -0.798 *** 0.813 *** [0.004] [0.001] [0.001] [0.000] DSCRT -0.758 *** -0.722 *** -0.837 *** 0.851 *** [0.002] [0.004] [0.000] [0.000] CPRT FEDRT DSCRT SPPE M2 EQUITY DEBT BAA CPRT 1 FEDRT 0.994 *** 1 [0.000] DSCRT 0.962 *** 0.972 *** 1 [0.000] [0.000] *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table IV. Correlations between Indexes of Conglomerate Return and Macro Variables: Bank-Dependent Compared to Bank-Independent Conglomerates EWCI, VWCI, and RI are the indexes of conglomerate return. I calculate EWCI (VWCI) as the logarithm of the ratio of the equal-weighted (value-weighted) average value changes between conglomerates and their matching firms; the value of the matching firm of each conglomerate as the sum of the imputed values of the conglomerate's segments as stand-alone firms; and RI as the logarithm of the ratio of the equal-weighted value changes between a portfolio of conglomerates and a portfolio of focused firms. FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated corporate bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues in each year, respectively. I define bank-independent conglomerates as those conglomerates with S&P bond ratings according to Compustat, and bank-dependent conglomerates as those conglomerates with no S&P bond ratings. SPPE M2 EQUITY Panel A. EWCI, An Equal-Weighted Index of Conglomerate Return EWCI (Bank-dependent) -0.948 *** -0.856 *** -0.684 *** EWCI (Bank-independent) -0.417 -0.202 -0.183 Panel B. VWCI, A Value-Weighted Index of Conglomerate Return VWCI (Bank-dependent) -0.906 *** -0.816 *** -0.679 *** VWCI (Bank-independent) -0.838 *** -0.755 *** -0.498 * Panel C. RI, An Equal-Weighted Index of Conglomerate Return RI (Bank-dependent) -0.938 *** -0.952 *** -0.632 ** RI (Bank-independent) -0.027 -0.041 -0.376 DEBT BAA CPRT Panel A. EWCI, An Equal-Weighted Index of Conglomerate Return EWCI (Bank-dependent) -0.710 *** 0.828 *** 0.636 ** EWCI (Bank-independent) 0.044 0.131 -0.032 Panel B. VWCI, A Value-Weighted Index of Conglomerate Return VWCI (Bank-dependent) -0.726 *** 0.840 *** 0.668 *** VWCI (Bank-independent) -0.503 * 0.691 0.535 ** Panel C. RI, An Equal-Weighted Index of Conglomerate Return RI (Bank-dependent) -0.748 *** 0.905 *** 0.731 *** RI (Bank-independent) 0.027 0.094 0.148 FEDRT DSCRT Panel A. EWCI, An Equal-Weighted Index of Conglomerate Return EWCI (Bank-dependent) 0.635 ** 0.609 ** EWCI (Bank-independent) -0.017 -0.138 Panel B. VWCI, A Value-Weighted Index of Conglomerate Return VWCI (Bank-dependent) 0.678 *** 0.669 *** VWCI (Bank-independent) 0.549 ** 0.493 * Panel C. RI, An Equal-Weighted Index of Conglomerate Return RI (Bank-dependent) 0.735 *** 0.729 *** RI (Bank-independent) 0.171 0.056 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table V. Conglomerate Excess Value and Capital Market Conditions I base all regressions on a sample of conglomerates. The dependent variable is the conglomerate excess value (E[V.sub.t]), calculated as the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. The independent variables include SIZE, the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC, operating income before interest and taxes deflated by sales revenue; CAPEX, capital expenditures deflated by sales revenue; and LTDR, the amount of long-term debt over the book value of assets. MKT is one of the following macro variables: FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues, respectively. The regressions results are based on 7,578 observations across 1,572 firms. (1) (2) (3) (4) SPPE M2 EQUITY DEBT Constant -0.528 *** 0.727 ** -0.381 *** -0.302 *** [0.000] [0.041] [0.000] [0.007] SIZ[E.sub.t-1] 0.083 *** 0.081 *** 0.077 ** 0.077 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t-1] -0.073 -0.068 -0.074 -0.07 [0.134] [0.164] [0.129] [0.148] CAPE[X.sub.t-1] 0.513 *** 0.509 *** 0.525 *** 0.519 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.668 *** 0.665 *** 0.680 *** 0.675 *** [0.000] [0.000] [0.000 [0.000] MK[T.sub.t] -0.006 *** -0.168 *** -0.042 *** -0.042 *** [0.000] [0.000] [0.000] [0.008] Chi Square 434.05 423.65 422.8 402.74 (5) (6) (7) (8) BAA CPRT FEDRT DSCRT Constant -0.748 *** -0.639 *** -0.637 *** -0.623 *** [0.000] [0.000] [0.000] [0.000] SIZ[E.sub.t-1] 0.080 *** 0.076 *** 0.076 *** 0.076 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t-1] -0.07 -0.071 -0.071 -0.07 [0.148] [0.145] [0.145] [0.150] CAPE[X.sub.t-1] 0.512 ** 0.519 *** 0.518 *** 0.519 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.665 *** 0.675 *** 0.674 *** 0.680 *** [0.000] [0.000] [0.000 * [0.000] MK[T.sub.t] 0.013 *** 0.007 ** 0.006 0.005 [0.001] [0.026] [0.022] [0.142] Chi Square 415.03 409.11 410.1 400.25 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0. 10 level. Table VI. Conglomerate Excess Value and Capital Market Conditions: Bank-Dependent Compared to Bank-Independent Conglomerates I base all regressions on a sample of conglomerates. The dependent variable is the conglomerate excess value (EV,), which is the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. The independent variables include SIZE, the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC, operating income before interest and taxes deflated by sales revenue; CAPEX, capital expenditures deflated by sales revenue; and LTDR, the amount of long-term debt over the book value of assets. MKT is one of the following macro variables: FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated bond rate; CPR * is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues, respectively. define as bank-dependent those firms with no S&P bond ratings, and as bank-independent those firms with S&P bond ratings. Bank-dependency equals one for bank-dependent firms and zero for bank-independent firms. The regression results are based on 6,566 observations across 1,452 firms. (1) (2) (3) (4) SPPE M2 EQUITY DEBT Constant -0.735 *** 0.682 -0.809 -0.806 *** [0.000] [0.262] [0.000] [0.000] SIZ[E.sub.t-1] 0.086 *** 0.084 ** 0.076 0.077 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t-1] -0.073 -0.064 -0.075 -0.074 [0.173] [0.228] [0.160] [0.167] CAPE[X.sub.t-1] 0.619 *** 0.612 *** 0.636 *** 0.625 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.691 *** 0.689 *** 0.714 *** 0.703 ** [0.000] [0.000] [0.000] [0.000] Bank-dependency 0.288 *** 0.708 0.701 ** 0.847 *** [0.000] [0.344] [0.000] [0.001] MK[T.sub.t](1- -0.005 *** -0.187 * 0.007 0.005 Bank-dependency) [0.004] [0.015] [0.702] [0.881] ([[gamma].sub.1]) MK[T.sub.t] -0.009 *** -0.247 *** -0.086 *** -0.085 *** Bank-dependency [0.000] [0.000] [0.000] [0.000] ([[gamma].sub.2]) P-value: [[gamma] 0.136 0.520 0.000 0.016 .sub.2]-[[gamma] .sub.1] Chi Square 309.16 299.43 311.94 285.45 (5) (6) (7) (8) BAA CPRT FEDRT DSCRT Constant -0.895 *** -0.745 *** -0.752 *** -0.724 *** [0.000] [0.000] [0.000] [0.000] SIZ[E.sub.t-1] 0.082 *** 0.076 *** 0.077 *** 0.076 [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t-1] -0.072 -0.074 -0.074 -0.073 [0.175] [0.169] [0.166] [0.171] CAPE[X.sub.t-1] 0.613 ** 0.619 *** 0.619 *** 0.622 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.687 *** 0.703 *** 0.702 *** 0.710 *** [0.000] [0.000] [0.000] [0.000] Bank-dependency 0.080 0.068 0.090 * 0.083 [0.400] [0.192] [0.063] [0.113] MK[T.sub.t](1- 0.01 -0.004 -0.003 -0.007 Bank-dependency) [0.210] [0.472] [0.534] [0.241] ([[gamma].sub.1]) MK[T.sub.t] 0.024 ** 0.019 ** 0.017 ** 0.016 ** Bank-dependency [0.000] [0.000] [0.000] [0.001] ([[gamma].sub.2]) P-value: [[gamma] 0.126 0.001 0.001 0.003 .sub.2]-[[gamma] .sub.1] Chi Square 295.15 292.46 292.06 281.78 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VII. Conglomerate Excess Value and Capital Market Conditions: An Alternative Specification with Differenced Variables I base all regressions on a sample of conglomerates. The dependent variable is the fast difference of conglomerate excess value [DELTA]E[V.sub.t]=E[V.sub.t]-E[V.sub.t-1], where EV is the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. The independent variables include SIZE, the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC, operating income before interest and taxes deflated by sales revenue; CAPER, capital expenditures deflated by sales revenue; and LTDR, the amount of long- term debt over the book value of assets. [DELTA]MK[T.sub.t] = MK[T.sub.t]-MK[T.sub.t-1]. MKT is one of the following macro variables: FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SHE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues, respectively. The regression results are based on 7,496 observations across 1,551 firms. (1) (2) (3) (4) [DELTA]SPPE [DELTA]M2 [DELTA]EQUITY [DELTA]DEBT Constant -0.063 *** -0.066 *** -0.067 *** -0.061 *** [0.000] [0.000] [0.000] [0.000] SIZ[E.sub.t1] 0.631 *** 0.630 *** 0.641 *** 0.644 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t1] 0.079 ** 0.082 ** 0.072 ** 0.080 ** [0.020] [0.016] [0.033] [0.017] CAPE[X.sub.t1] 0.347 *** 0.343 *** 0.325 *** 0.331 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t1] -0.561 *** -0.568 *** -0.610 *** -0.602 *** [0.000] [0.000] [0.000] [0.000] [DELTA]MK -0.005 *** -0.469 ** -0.055 *** -0.102 *** [T.sub.t1] [0.005] [0.001] [0.000] [0.000] R Square 0.330 0.331 0.336 0.339 (5) (6) (7) (8) [DELTA]BAA [DELTA]CPRT [DELTA]FEDRT [DELTA]DSCRT Constant -0.061 *** -0.070 *** -0.071 *** -0.072 *** [0.000] [0.000] [0.000] [0.000] SIZ[E.sub.t1] 0.637 *** 0.629 *** 0.627 *** 0.626 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t1] 0.081 ** 0.077 ** 0.078 ** 0.079 ** [0.017] [0.023] [0.021] [0.020] CAPE[X.sub.t1] 0.341 *** 0.344 *** 0.349 *** 0.351 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t1] -0.596 *** -0.570 *** -0.555 *** -0.552 *** [0.000] [0.000] [0.000] [0.000] [DELTA]MK 0.023 *** 0.006 ** 0.001 0 [T.sub.t1] [0.000] [0.012] [0.570] [0.979] R Square 0.333 0.330 0.329 0.329 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table VIII. Differenced Conglomerate Excess Value and Capital Market Conditions: Bank-Dependent Compared to Bank-Independent Conglomerates I base all regressions on a sample of conglomerates. The dependent variable is the first difference of conglomerate excess value [DELTA]E[V.sub.t] = E[V.sub.t]-E[V.sub.t-1], where EV is the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. The independent variables include SIZE, the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC, operating income before interest and taxes deflated by sales revenue; CAPEX, capital expenditures deflated by sales revenue; and LTDR, the amount of long-term debt over the book value of assets. [DELTA]MK[T.sub.t] = MK[T.sub.t]-MK[T.sub.t-1]. MKT is one of the following macro variables: FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues, respectively. I define as bank-dependent those firms with no S&P bond ratings, and as bank-independent those firms with S&P bonding ratings. Bank-dependency equals one for bank-dependent firms and zero for bank-independent firms. The regression results are based on 6,506 observations across 1,433 firms. (1) (2) (3) [DELTA]SPPE [DELTA]M2 [DELTA]EQUITY Constant -0.062 *** -0.069 *** -0.068 *** [0.000] [0.000] [0.000] SIZ[E.sub.t-1] 0.662 *** 0.658 *** 0.672 *** [0.000] [0.000] [0.000] LTD[R.sub.t-1] 0.061 0.063 0.052 [0.112] [0.101] [0.172] CAPE[X.sub.t-1] 0.309 *** 0.305 *** 0.278 *** [0.000] [0.000] [0.000] OPIN[C.sub.t-1] -0.634 *** -0.633 *** -0.669 *** [0.000] [0.000] [0.000] [DELTA]MK[T.sub.t] -0.005 0.162 -0.005 * (1-Bank-dependency) [0.121] [0.520] [0.680] ([[gamma].sub.1]) [DELTA]MK[T.sub.t] -0.008 *** -0.571 *** -0.089 *** * Bank-dependency [0.005] [0.006] [0.000] ([[gamma].sub.2]) P-value: [[gamma].sub.2] - 0.466 0.022 0.000 [[gamma].sub.1] R Square 0.331 0.331 0.340 (4) (5) (6) [DELTA]DEBT [DELTA]BAA [DELTA]CPRT Constant -0.063 *** -0.060 *** -0.070 *** [0.000] [0.000] [0.000] SIZ[E.sub.t-1] 0.672 *** 0.668 *** 0.658 *** [0.000] [0.000] [0.000] LTD[R.sub.t-1] 0.062 0.062 0.06 [0.103] [0.107] [0.119] CAPE[X.sub.t-1] 0.288 *** 0.297 *** 0.303 *** [0.000] [0.000] [0.000] OPIN[C.sub.t-1] -0.662 *** -0.665 *** -0.630 *** [0.000] [0.000] [0.000] [DELTA]MK[T.sub.t] -0.015 0.002 -0.009 ** * (1-Bank-dependency) [0.439] [0.724] [0.042] ([[gamma].sub.1]) [DELTA]MK[T.sub.t] -0.146 *** 0.041 *** 0.014 *** * Bank-dependency [0.000] [0.000] [0.000] ([[gamma].sub.2]) P-value: [[gamma].sub.2] - 0.000 0.000 0.000 [[gamma].sub.1] R Square 0.340 0.337 0.333 (7) (8) [DELTA]FEDRT [DELTA]DSCRT Constant -0.072 *** -0.073 *** [0.000] [0.000] SIZ[E.sub.t-1] 0.657 *** 0.657 *** [0.000] [0.000] LTD[R.sub.t-1] 0.062 0.06 [0.109] [0.117] CAPE[X.sub.t-1] 0.310 ** 0.316 *** [0.000] [0.000] OPIN[C.sub.t-1] -0.620 *** -0.617 *** [0.000] [0.000] [DELTA]MK[T.sub.t] -0.011 *** -0.023 *** * (1-Bank-dependency) [0.005] [0.000] ([[gamma].sub.1]) [DELTA]MK[T.sub.t] 0.007 ** 0.009 ** * Bank-dependency [0.024] [0.035] ([[gamma].sub.2]) P-value: [[gamma].sub.2] - 0.000 0.000 [[gamma].sub.1] R Square 0.332 0.333 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table IX. Conglomerate Excess Value and Capital Market Conditions: Large Compared to Small Conglomerates I base all regressions on a sample of conglomerates. The dependent variable is conglomerate excess value (E[V.sub.t]), which is the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. The independent variables include SIZE, the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC, operating income before interest and taxes deflated by sales revenue; CAPEX, capital expenditures deflated by sales revenue; and LTDR, the amount of long-term debt over the book value of assets. MKT is one of the following macro variables: FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-eamings ratio of the S&P 500; BAA is the BAA rated bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues, respectively. Dummy for large size equals one for those conglomerates with SIZ[E.sub.t-1] above the median SIZ[E.sub.t-1] of all conglomerates; and dummy for small size equals one for those conglomerates with SIZ[E.sub.t-1] below the median SIZ[E.sub.t-1] of all conglomerates. The regression results are based on 7,578 observations across 1,572 firms. (1) (2) (3) SPPE M2 EQUITY Constant -0.629 *** -0.426 -0.506 *** [0.000] [0.369] [0.000] SIZ[E.sub.t-1] 0.088 *** 0.085 *** 0.082 *** [0.000] [0.000] [0.000] LTD[R.sub.t-1] -0.080 * -0.076 -0.075 [0.098] [0.114] [0.121] CAPE[X.sub.t-1] 0.514 *** 0.513 *** 0.525 *** [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.661 *** 0.657 *** 0.684 *** [0.000] [0.000] [0.000 -0.161 *** -2.319 *** -0.199 Dummy for Small Size [0.001] [0.000] [0.068] MK[T.sub.t] * Dummy for -0.003 * -0.029 -0.026 * Large Size ([[gamma].sub.l]) [0.082] [0.624] [0.052] MK[T.sub.t] * Dummy for -0.010 *** -0.314 *** -0.060 *** Small Size ([[gamma].sub.2]) [0.000] [0.000] [0.000] P-value:[[gamma].sub.2] - 0.001 0.001 0.107 [[gamma].sub.1] Chi Square 452.63 439.18 432.59 (4) (5) (6) DEBT BAA CPRT Constant -0.626 *** -0.669 *** -0.616 *** [0.000] [0.000] [0.000] SIZ[E.sub.t-1] 0.082 *** 0.084 *** 0.080 *** [0.000] [0.000] [0.000] LTD[R.sub.t-1] -0.075 -0.079 -0.077 [0.122] [0.103] [0.112] CAPE[X.sub.t-1] 0.521 *** 0.515 *** 0.520 *** [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.674 *** 0.661 *** 0.682 *** [0.000] [0.000] [0.000] -0.597 *** 0.212 *** 0.101 ** Dummy for Small Size [0.006] [0.008] [0.025] MK[T.sub.t] * Dummy for -0.002 0.002 -0.003 Large Size ([[gamma].sub.l]) [0.923] [0.788] [0.481] MK[T.sub.t] * Dummy for -0.083 *** 0.026 *** 0.017 *** Small Size ([[gamma].sub.2]) [0.000] [0.000] [0.000] P-value:[[gamma].sub.2] - 0.009 0.001 0.001 [[gamma].sub.1] Chi Square 412.05 428.02 423.67 (7) (8) FEDRT DSCRT Constant -0.620 *** -0.598 *** [0.000] [0.000] SIZ[E.sub.t-1] 0.081 *** 0.080 *** [0.000] [0.000] LTD[R.sub.t-1] -0.077 -0.077 [0.111] [0.112] CAPE[X.sub.t-1] 0.519 *** 0.521 *** [0.000] [0.000] OPIN[C.sub.t-1] 0.680 *** 0.684 *** [0.000] [0.000] 0.088 ** 0.103 ** Dummy for Small Size [0.035] [0.022] MK[T.sub.t] * Dummy for -0.002 -0.006 Large Size ([[gamma].sub.l]) [0.510] [0.194] MK[T.sub.t] * Dummy for 0.015 *** 0.017 *** Small Size ([[gamma].sub.2]) [0.000] [0.001] P-value:[[gamma].sub.2] - 0.000 0.000 [[gamma].sub.1] Chi Square 425.35 412.93 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level. Table X. Conglomerate Excess Value and Capital Market Conditions: High KZ Compared to Low KZ Conglomerates I base all regressions on a sample of conglomerates. The dependent variable is conglomerate excess value (E[V.sub.t]), which is the logarithm of the ratio of a conglomerate's market value over the sum of the imputed values of the conglomerate's segments as stand-alone firms. The independent variables include SIZE, the logarithm of market value, where market value is the book value of assets minus the book value of equity plus the market value of equity; OPINC, operating income before interest and taxes deflated by sales revenue; CAPEX, capital expenditures deflated by sales revenue; and LTDR, the amount of long-term debt over the book value of assets. MKT is one of the following macro variables: FEDRT is the Federal Funds rate; DSCRT is the discount rate charged by the Federal Reserve; M2 is the logarithm of the M2 money supply; SPPE is the price-to-earnings ratio of the S&P 500; BAA is the BAA rated bond rate; CPRT is the six-month commercial paper rate; and EQUITY and DEBT are the logarithms of the amounts of equity issues and debt issues, respectively. I calculate Kaplan and Zingales (1997) index (KZ index) using the coefficients in Lamont et al. (2001). The high KZ dummy equals one for those conglomerates with K[Z.sub.t] above the median K[Z.sub.t] of all conglomerates; and the low KZ dummy equals one for those conglomerates with K[Z.sub.t] below the median K[Z.sub.t] of all conglomerates. The regression results are based on 7,299 observations across 1,530 firms. (1) (2) (3) (4) SPPE M2 EQUITY DEBT Constant -0.539 *** 1.135 ** -0.330 *** -0.233 [0.000] [0.018] [0.000] [0.140] SIZ[E.sub.t-1] 0.076 *** 0.074 *** 0.071 *** 0.071 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t-1] 0.027 0.033 0.032 0.036 [0.612] [0.536] [0.554] [0.507] CAPE[X.sub.t-1] 0.597 *** 0.591 *** 0.608 *** 0.602 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.695 *** 0.694 *** 0.696 *** 0.696 *** [0.000] [0.000] [0.000] [0.000] The High KZ Dummy 0.001 -1.219 ** -0.155 -0.279 [0.970] [0.046] [0.132] [0.179] MK[T.sub.t] * The -0.003 * -0.061 -0.013 -0.006 Low KZ Dummy ([[gamma].sub.1]) [0.083] [0.280] [0.333] [0.769] MK[T.sub.t] * The -0.007 *** -0.224 *** -0.061 *** -0.059 *** High KZ Dummy ([[gamma].sub.2]) [0.000] [0.000] [0.000] [0.008] P-value: 0.017 0.032 0.016 0.073 [[gamma].sub.2] - [[gamma].sub.1] Chi Square 499.83 489.54 505.25 478.22 (5) (6) (7) (8) BAA CPRT FEDRT DSCRT Constant -0.829 *** -0.709 *** -0.700 *** -0.673 *** [0.000] [0.000] [0.000] [0.000 SIZ[E.sub.t-1] 0.073 *** 0.070 *** 0.070 *** 0.069 *** [0.000] [0.000] [0.000] [0.000] LTD[R.sub.t-1] 0.033 0.035 0.035 0.037 [0.542] [0.511] [0.510] [0.485] CAPE[X.sub.t-1] 0.594 *** 0.600 *** 0.600 *** 0.602 *** [0.000] [0.000] [0.000] [0.000] OPIN[C.sub.t-1] 0.691 ** 0.697 *** 0.696 *** 0.703 *** [0.000] [0.000] [0.000] [0.000] The High KZ Dummy 0.226 *** 0.159 *** 0.143 *** 0.123 *** [0.001] [0.000] [0.000] [0.001] MK[T.sub.t] * The 0.004 0 0.001 0 Low KZ Dummy ([[gamma].sub.1]) [0.463] [0.982] [0.813] [0.932] MK[T.sub.t] * The 0.018 *** 0.010 ** 0.009 ** 0.006 High KZ Dummy ([[gamma].sub.2]) [0.001] [0.015] [0.022] [0.242] P-value: 0.051 0.062 0.108 0.400 [[gamma].sub.2] - [[gamma].sub.1] Chi Square 485.26 483.21 482.52 471.22 *** Significant at the 0.01 level. ** Significant at the 0.05 level. * Significant at the 0.10 level.

Printer friendly Cite/link Email Feedback | |

Author: | Yan, An |
---|---|

Publication: | Financial Management |

Geographic Code: | 1USA |

Date: | Dec 22, 2006 |

Words: | 14553 |

Previous Article: | Corporate social performance and stock returns: UK evidence from disaggregate measures. |

Next Article: | What drives the S&P 500 inclusion effect? An analytical survey. |

Topics: |