While the Indian manufacturing sector has experienced rapid growth since the early 1990s, it is characterised by large productivity differences across firms and presence of several low productivity firms that use poor managerial practices. This column examines differences in CEOs’ management style via their time use to provide new insights on the observed diversity across firms.
Management practices play a key role in explaining the differences in productivity across and within countries, and yet little is known about the top executives who shape these practices and who have a large impact on firm performance.
What do CEOs actually do? Does it differ across firms? Is it correlated with firm performance? If so, why do different CEOs behave differently?
Measuring CEO time use
To provide answers, we develop a new survey instrument to measure CEO time use and employ it to collect data on the time use of 354 CEOs of listed Indian manufacturing firms over the course of one randomly chosen work week (Bandiera et al 2013). India is an interesting context for this exercise not just because of its size and relative poverty, but also because its manufacturing sector has experienced a burst of rapid growth since the early 1990s but is nevertheless characterised by large productivity differences across firms and presence of several low productivity firms that use poor managerial practices.
To measure time use we reconstruct the CEOs time diary via daily phone interviews with their personal assistants. We ask respondents to use their actual diaries to list sequentially all activities that lasted longer than fifteen minutes, and for each activity we enquire about its type (meeting, phone call etc.), the type and number of people involved, the location, the start and end time, and scheduling horizon. This allows us to build an accurate bottom-up estimate of the CEO labour supply and its allocation across different activities.
We complement the time use data with information about the CEOs’ and firms characteristics and we match these to external balance sheet data to assess the extent to which time use vary across CEOs, whether this is correlated with firm performance and if so, what drives these differences. The analysis yields three sets of findings.
Large differences in labour supply and management style
First, CEOs differ substantially both on their labour supply and management style. The median1 CEO in our sample works 7 hours a day (net of commuting and personal activities), but labour supply is just 6 hours a day at the 25th percentile2 and 8 hours a day at the 75th percentile.
We also find that CEOs differ remarkably in their style, namely their time allocation for a given number of total hours worked. In particular, CEOs differ in the extent they plan in advance, the time they devote to face-to-face meetings, the composition of people at those meetings, the time devoted to outsiders vs. employees of the firm, and, among these, the time devoted to direct reports3.
Our analysis reveals that CEOs in our sample display two distinct patterns of time allocation, which we label “Styles”. Style 1 CEOs show a higher tendency to engage in activities that are planned in advance, have a multi-participant and cross-functional component, are directed at their own employees, especially in production, and involve exclusively their direct reports. In contrast, Style 2 CEOs spend more time in unplanned activities, meet fewer people in any given interaction, and are more likely to interact with outsiders rather than firm employees. In our sample, 37% of CEOs adopt Style 1.
Differences in time use are linked to firm productivity and profitability
Second, these differences in labour supply and style are strongly linked with external measures of firm productivity and profitability: a 1% increase in the weekly hours worked by the CEO is associated with a 0.89% increase in firm productivity, and firms led by Style 1 CEOs are on average 34% more productive4.
The external environment does not affect time use
Third we show that the external environment in which the firm operates, namely the specific industry and a several measure of state policies and infrastructure, does not explain much of the variation in time use we observe in the data. Instead, we find that family CEOs - executives belonging to the family who owns the firm - work 7% fewer hours than other managers and have a 13% lower probability of adopting Style 1. These findings are in line with evidence showing that family CEOs typically underperform and that they adopt worse managerial practices.
Family and professional CEOs use their time differently
To conclude, we investigate whether the difference between family and professional CEOs is sensitive to the level of competitive pressure the firm is exposed to. Finding that exposure to competition reduces the difference would lend support to the idea that this depends at least in part on the different set of incentives faced by the CEO.
In line with this, we find that the differences in labour supply and style are driven entirely by family CEOs employed in firms that are less exposed to foreign or domestic competition.
We also exploit a natural experiment, variation in monsoon rainfall within the survey week, to compare the change in time use of family CEOs to that of professional CEOs. Under the assumption that the monsoon rainfall increases the cost of working for all CEOs equally regardless of firm ownership (e.g. because going to work takes much longer), this strategy allows us to shed light whether family and professional managers put different weight on firm performance. Extreme rainfall is associated with a 10% decrease in daily hours worked for family CEOs, but has no significant impact on other CEOs. Moreover, we show that the differential reaction to the external shocks is entirely driven by family CEOs who lead firms that are less exposed to competition. Taken together these findings suggest that differences between family and professional CEOs are due to differences in preferences or incentives rather than optimal responses to different organisational structures.
CEO management style matters
Overall, the observed variation in CEO style is in line with existing evidence of large dispersion in productivity and managerial practices among Indian manufacturing firms. This naturally raises the questions of whether a style 1 is associated with better performance in different contexts and if the process of development leads to convergence on a pattern of time use dominated by style 1 and long working hours.
- The median of a list of numbers can be found by arranging all the observations from lowest value to highest value and picking the middle one.
- A percentile is the value of a variable below which a certain percent of observations fall. Here, the 25th percentile is the value below which 25 percent of the observations may be found.
- To combine these differences in a single measure we use standard clustering procedures.
- While we do not observe exogenous variation in time use that would allow us to establish its causal impact on performance, a placebo test reassuringly shows that the observed correlation is not driven by time invariant firm traits. Indeed, firm performance prior to the appointment of the current CEO, which captures time invariant firm traits, is not correlated with the labour supply and style of the CEO during his tenure.
- Bandiera, Oriana, Prat, Andrea and Raffaella, Sadun (2013) “Managing Firms In An Emerging Economy: Evidence From The Time Use Of Indian Ceos”, mimeo LSE
- Bandiera, Oriana, Guiso, Luigi, Prat, Andrea and Sadun, Raffaella (2011) "What Do CEOs Do?," CEPR Discussion Papers 8235, CEPR Discussion Papers