Regulation, Ownership, and Costs

In our paper, Regulation, Ownership, and Costs: A Historical Perspective from Indian Railways, forthcoming in the AEJ: Economic Policy, we provide an historical perspective by studying the transition from private to colonial state ownership of Indian railways from 1874 to 1912. In the mid-19th century prompted by British merchant houses and railway promoters, the British Government in London encouraged railway development in India. It opted for a system of private British owned and operated railways. However, the contracts stipulated a 5 percent dividend guarantee on share capital payable by the colonial Government of India. Such guarantees were common in other countries and were designed to compensate British investors for the risk involved in building railways in foreign places (Eichengreen 1995).

On account of conflicts and decades of disappointing performance, the Government of India began to construct and operate state lines in the 1870s. At the same time, they also began to takeover private companies. Because of a clause in the original concession contract, the Government could only takeover private companies on either the 25^th or 50^th anniversary of their contract. The Government exercised the takeover option in every case and by 1910 formed an extensive ownership stake in the railway sector. But, this process did not eliminate the private sector. Many companies were allowed to retain operations, but they faced more stringent Government control and supervision.

Our empirical analysis studies whether this change in ownership and regulatory structure influenced operating costs. We focus on costs because it is a key performance measure in the theoretical literature on ownership and regulation. Broadly the literature emphasizes how different objectives, institutional arrangements, and incentive structures affect managerial efforts to reduce marginal costs (see Estache and Wren-Lewis 2009 for review). Thus our analysis speaks to the cost of delivering infrastructure services once fixed network investments have been made.

Using administrative reports from the Government of India, we constructed a new historical data set on the major railway systems operating in India. The data include detailed information on mileage, passenger and goods traffic, fuel prices, wages, costs, accidents, and ownership status from 1874 to 1912. In terms of costs, the data distinguish between working expenses and capital expenditures. Working expenses represents the cost of hiring train and station staff, fuel costs, maintenance expenditures, and administration. Capital expenditure represents the value of track miles, locomotives, vehicles, and stations.

We begin by estimating a variable cost function where the dependent variable is working expenses for a railway system in a given year. Our right hand side includes common variables in a cost function such as output, fixed capital, and input prices along with a dummy for state ownership. The state dummy takes the value 0 under the initial regime where there was private ownership and operation with high guarantees and weak Government oversight. It takes the value 1 under the subsequent regime where railway systems had majority Government of India ownership with Government operation or private operation with lower guarantees and stronger Government oversight. We also include railway fixed effects, year fixed effects, and railway specific trends to control for unobservable factors trending up or down that may contaminate the estimates on state ownership. Our results show that variable costs declined under Government of India ownership compared to private British ownership. Working expenses were 14 percent lower on average following a move to state ownership and the accompanying regulatory changes.

Unlike traditional cross-sectional comparisons, our institutional setting mitigates endogeneity problems of selection bias and omitted variables. There is no selection problem because the Government took over all the original private companies. In addition, the contractual environment ensured the Government could not endogenously time takeovers to coincide with periods of increasing or decreasing costs. But, private companies could perhaps foresee organizational changes ahead of their contract deadlines and take actions to increase or decrease variable costs in anticipation of a state takeover (i.e. anticipation effects). To investigate such dynamics, we construct a sequence of dummy variables for intervals before and after state takeovers. But, we find no evidence of a rise in costs either 3 or 5 years before takeovers. We find no heterogeneous anticipation effects between early switchers where the probability of takeover was more uncertain relative to post-1900 switchers where there was less uncertainty about a takeover.

We also find no evidence of the Government lowering maintenance expenditures shortly after the takeover to boost short-run profits while raising long-run costs through a dilapidated capital stock. If anything, we find the big decline in working expenses occurred 11 to 15 years after takeover and continued for 20 years. Moreover, the cost reductions did not come at the expense of quality. We observe no change in safety measured by the frequency of accidental injuries or deaths following takeovers.

To better understand the mechanisms underlying our findings, we replicate the regressions using subcategories of working expenses as the dependent variable. The subcategories include working expenses in the traffic, maintenance, locomotive, vehicle, and administrative departments separately. We find the shift to state ownership decreased working expenses in traffic and administrative departments, which used more labor. Complementing these results, we also observe the total number of workers, including Europeans, declined after takeovers. Thus, it appears the Government of India reduced costs by cutting labor including their more expensive workers.

The full paper is available for download here.