DO LOCAL MANAGERS GIVE LABOR AN EDGE?
4.4.2 How Do Local Managers Finance Higher Employment
The results to this point show that local managers are less likely to lay off workers during times of industry distress than are their non-local peers. It is interesting to ask how this higher level of employment is funded. In this section I investigate several possible ways that local managers could finance these relatively higher employment levels. Specifically, I ask whether following industry downturns local managers are more likely to: reduce payouts to shareholders, sell assets, reduce investment spending, spend cash, or increase debt.
Payout policy
One way managers could fund higher employment is by cutting direct transfers to shareholders. This action would be an immediate direct transfer of wealth from shareholders to employees. I investigate changes in payout policy around industry distress for this reason. I do so by estimating Equation (4.1) with different measures of industry-adjusted shareholder payout as the dependent variable. Specifically I look at the frequency of various levels of share repurchases and of dividend payments.
If managers cut payments to shareholders to pay for increased employment during industry downturns, then the estimate of βL should be negative and significant. Since I am interested in changes in payout policy, I include both firm fixed effects and the one year lagged payout variable. The persistence of payout policy is captured by the lagged variable and the firm’s average payout policy is controlled for with the firm fixed effects. In addition, I include industry-adjusted control variables to control for: size, past operating and stock market performance, stock market volatility, cash holdings, and leverage.
Table 4.8 shows the regression results. Columns (1) through (5) display the results for repurchases of various sizes relative to the book value of assets. For all levels of repurchases-to-assets there is no difference in the change in repurchase payments for firms run by local managers versus those run by non-local managers during industry downturns. The estimate ofβLin all cases is not reliably different from zero. Dividend payments are explored in column (6). The estimate of βL in column (6) is also indistinguishable from zero. As was the case for repurchases, dividend payments to shareholders are not more likely to be cut in firms run by local managers than those run by non-local managers following industry distress. The evidence from Table 4.8 indicates that local managers do not fund higher employment during industry distress by decreasing payouts to shareholders.
Asset Sales
Atanassov and Kim (2009) show that managers in countries with weak shareholder protections and strong labor unions often sell firm assets to avoid layoffs in hopes of gaining laborers as allies. It is possible that the same may be true for local managers. Local managers may fund higher levels of employment by selling off firm assets. To test this theory I use the same empirical methodology as before. I estimate Equation (4.1) with the dependent variable equal to an industry-adjusted dummy variable indicating whether the firm sold more than a specified percentage of their total assets. The levels of asset sales-to-assets that I investigate are greater than 0, 1, 5, and 10 percent. If local managers are more likely to sell assets than their non-local counterparts during times of industry distress, then the estimate of βL should be positive. I include industry-adjusted firm-specific controls variables to control for:
firm size, past operating and stock market performance, stock market volatility, cash holdings, and leverage.
The regression results are displayed in Table 4.9. Under all four definitions of asset sales I find that the estimate ofβL is not statistically different from zero, indicating that during times of industry distress asset sales are no different between firms run by local managers versus those run by non-locals. Given that layoffs in Atanassov and Kim (2009) are defined as reductions in the workforce of at least 20 percent, it is not that surprising that the results in this paper are different. The previous results show that local managers are only able to influence smaller reductions in employment. It may not be necessary to sell assets avoid the intermediate size workforce reductions that local managers are able to reduce.
Changes in Investment, Cash, and Leverage
Thus far I have shown that local managers do not fund their relative increases in employment levels during industry downturns by either cutting payments to share- holders or selling off assets. I next explore three other ways that managers may fund these relatively higher levels of employment. Specifically, I look at changes in investment spending, cash holdings, and leverage. I test the changes in these variables by estimating Equation (4.1) including firm fixed effects and the lagged dependent variable. The dependent variables I investigate are industry-adjusted investment, cash, and leverage and are all scaled by the book value of total assets. I estimate each of these regressions first omitting firm-specific industry-adjusted control variables and then including a baseline specification of control variables based on recent work in these areas of the literature. Specifically the baseline models of investment, cash holdings, and leverage are based on specifications in Duchin, Ozbas, and Sensoy (2010), Bates, Kahle, and Stulz (2009), and Frank and Goyal (2007), respectively.72 If local managers reduce investment spending or cash holdings relatively more than non-locals during
72Note that some of the variable definitions differ slightly from those in cited papers. Detailed definitions of all variables are found in the appendix.
industry downturns, then the estimate of βL should be negative. If local managers are more likely to raise debt during industry downturns than non-locals, then the estimate of βL should be positive.
Table 4.10 reports the regression results testing the outlined hypotheses. The results in columns (1) and (2) indicate that local managers do not differentially change investment during times of poor industry performance from their non-local industry peers. In both specifications the estimate of βL is not significantly different from zero.
In columns (3) and (4), I investigate whether relative cash holdings within industries change differently following poor performance for local managers compared to non- locals. The estimate of βL in column (3) is -0.0096 and is significant at the 5 percent level. This estimate suggests that the cash holdings of firms run by local managers fall by about one percent of assets relative to their non-local industry peers during times of industry distress. When controlling for within industry differences in firm specific variables (column (4)) the estimate of βL is -0.0099 and is significant at the 10 percent significance level. Given that the average firm in the sample holds 14.8%
of its assets in cash, a one percent reduction in the cash-to-assets ratio of the firm is an economically significant 6.8% reduction in cash holdings.
Asking whether firms run by local managers take on debt to finance higher employment levels during industry downturns, I estimate the regressions displayed in columns (5) and (6) of the table. In both specifications the estimate of βL is not significantly different from zero. This implies that firms run by local versus non-local managers do not make different decisions regarding debt financing during industry downturns. Specifically, we reject the hypothesis that local managers raise more debt in relation to non-locals to fund higher levels of employment.
The evidence from this section shows that following periods of poor industry performance, the cash-to-assets ratio of firms run by local managers falls by roughly
one percent relative to the cash-to-assets ratio of firms run by their non-local industry counterparts. Earlier evidence shows that during these same downturns, the employ- ment growth rates of firms run by local managers are on average 0.0336 (Table 4.3 column (6)) higher than they are for firms run by non-local managers within the same industry. It is interesting to ask whether this relative reduction in cash holdings could feasibly fund the observed relative increase in workers. To judge whether this is feasible, let’s take the median firm in the sample during 2003 (the year with the most distressed observations). This firm had assets of $1,063 million and 5,300 employees in 2002. The estimates from the baseline model on employment growth suggest that if this firm were managed by a local CEO the firm would have 178 (0.0336×5,300) more workers and $10.524 million ($1,063 million × 0.0099) less cash at the end of 2003 than if it were run by a non-local manager. If the entire difference in cash is used to fund the higher level of employment, then this implies that the average cost per worker for the firm is $59,123 ($10.524 million / 178). Given that the American Community Survey reports that the median household income in the U.S. in 2003 was roughly
$43,500 and this does not include additional employee costs, such as employment taxes, health benefits, and retirement benefits, the reduction in relative cash holdings seems to be of a reasonable magnitude to fund the additional employment levels attributed to local managers.