New Institutional Economics : Understanding transaction costs
Revisiting one of the central concepts in the NIE+ framework — the theory of transaction costs and highlighting the methods of measuring such costs
IIn the past two years, I have discussed various aspects of New Institutional Economics (NIE) and how it diverges from Neoclassical economics. I integrated the concept of social capital with NIE and called it the NIE+ framework. I applied this theoretical framework to analyze various important public policy issues, which are important across the world. Using the framework, I also attempted to answer the question: why some countries grow and others fall by the wayside. I did so by looking at the development experience of various European countries from the 15th-19th centuries.
We may recall that the central concepts in the NIE+ framework are: transaction costs, asymmetric and imperfect information, collective action dilemmas in the supply of public goods/common property resources, property rights, institutions and social capital. These concepts have been discussed and applied in the various articles I have presented here. However, we need to be able to measure these concepts for a better analysis. Measurement will also enrich the analysis by allowing us to take up statistical modelling and explore correlations and possible causations. In this article, we will revisit the concept of transaction costs and look at how transaction costs can be measured.
REVISITING 'Transaction costs'
The concept of transaction cost was probably used first by Ronald Coase in his paper, The Nature of the Firm. We may recall that Coase was asking a very basic question: why do firms exist. He went on to develop a theoretical framework wherein he concluded that firms exist because they can do many of the economic tasks such as organising production, more efficiently than markets. He suggested that there were costs of using the markets, which he referred to as "costs of using the price mechanism". In his paper, The Problem of Social Cost, from which the famous Coase Theorem originated, he referred to the costs of market transactions while discussing a solution to externalities. In short, two types of costs are typical in an economy: costs of production and distribution and costs of exchange and transactions. And the firm's objective is to organise production in a manner that minimises transaction costs.
However, it was perhaps Oliver Williamson, the Nobel Prize winner in 2009, who popularised the field of study known today as 'Transaction Cost Economics'. Williamson took forward Coase's ideas and applied them to organisation theory. According to him, transaction costs determine the alternative forms of economic organisations and contractual arrangements. He dug deeper into Coase's theory and explained the boundaries of the firm and why some transactions occurred within the firm and others in the marketplace: the reason was basically these minimised transaction costs. For Williamson, it was the relative ranking of transaction costs in various organisations that held the key. In most empirical studies in this tradition, transaction costs are not measured directly, but proxy variables such as uncertainty (since contracts are generally incomplete and can't be fully specified ex-ante), asset specificity (the asset supplied is highly specialised and specific to the need of the buyer), the opportunity cost of doing something within the firm vis-à-vis the market etc. are used.
Milgrom and Roberts in their paper, Perspectives on Positive Political Economy (1990), added two costs to the theory proposed by Coase and developed by Williamson: bargaining costs, which are the short term costs of using the market to carry out transactions (rather than bringing them within a centralised structure like that of the firm) and influence costs, which are the costs of using a centralised authority like the firm as opposed to the market. This was in contrast to Williamson's theory which emphasised asset specificity, frequency of dealings and uncertainty. Further, while Williamson's theory is clear on the dichotomy of the firm and the market, it is on weaker ground when it claims that transacting through the hierarchy of the firm reduces transaction costs as opposed to operating in the marketplace.
While formal definitions of the phrase 'transaction cost' have been suggested by many economists, there is no consensus. Steven Cheung defines transaction costs as any costs that don't exist in a "Robinson Crusoe economy" — in other words, any costs that arise due to the existence of the market and institutions. Other economists restrict the definition to exclude costs internal to an organisation. In the neoclassical world, the Walrasian model with no institutions is the one with zero transaction costs.
Douglass North, the Nobel Prize winner in 1993, suggested that institutions, which are 'a set of rules that constrain behaviour', are key in determining transaction costs. Those institutions that minimise transaction costs also lead to faster economic growth.
Measuring transaction costs
There are very few studies that have set out to measure transaction costs in the economy or a sector of the economy. Wallis and North's study on measuring transaction costs in the US was noteworthy. However, the study wasn't really measuring transaction costs. Wallis and North basically measured what they defined as the 'transaction sector'. According to them, the whole economy can be divided into two main sectors: the production or transformation sector and the transaction sector. They measured the total value of resources used in the transaction sector and called this the transaction costs in the economy. In their study, they showed that the proportion of the US GNP in the transaction sector grew from 25 per cent in 1870 to 45 per cent in 1970. The more developed the economy, the bigger is the transaction sector. This was obviously not the same as transaction costs. To quote from their paper of 1986:
"While as economists we wish to separate transaction costs from other costs, individual economic actors have no such motivation. People maximise net benefits, the difference between total benefits and total costs, where total costs include both transaction and other costs. Every economic activity involves elements of transaction and other costs. Ideally our measure of transaction costs would delve into each exchange and separate these costs. Unfortunately, data are not available for such a measure. Instead, our basic approach is to segregate economic activities and actors into those that are primarily associated with making exchanges and those that are not. The sum of these sources used by those associated with transacting make up our estimate of the transaction sector."
The study by Wallis and North was problematic since it did not measure transaction costs as understood in the NIE theory, propounded by Coase and Williamson. Transaction costs in theory are something that lower efficiency and therefore should be minimised. This is contrary to what Wallis and North found: that the bigger the transaction sector, more developed the economy.
Another study that measured transaction costs was the one by Hernando Soto on the costs of doing business in a developing country like Peru. In the study titled The Other Path, he found that the formal legal system actually raised the costs of doing business. Hernando Soto's study focussed on the non-market transaction costs such as resources spent in getting permits and licenses, red-tape, money spent on bribery etc. In fact, the formal legal system was so obstructive that it has given rise to a large informal sector unprotected by formal law.
Stone, Levy and Paredes also attempted to measure transaction costs in the garment industry in Brazil and Chile and found that formal and legal regulatory systems are not always more efficient than informal systems. The Chilean system does indeed have legal simplicity and transparency and consistency of enforcement, and the businesses do benefit from this system. On the other hand, even though the Brazilian regulatory system is far more complex than Chile's, businesses in Brazil found more efficient practical substitutes to formal laws. The finding is similar to Hernando Soto's in that it "…warns against a preoccupation with formal legal and regulatory reform as an immediate means to promote economic development".
Where do we go from here
It is clear from the above discussion that measuring transaction costs is a work in progress. While the theory is well developed, few empirical studies have been undertaken. However, there is no single unified research methodology and I think that empirical researchers are still finding their way. I would imagine that any empirical study would differ from sector to sector and would vary across countries. For example, in a sector such as finance, where large amounts of data are available and figures on variables such as interest rates, brokerage fees etc. are readily available, it is easier to estimate transaction costs. On the other hand, in the field of environmental economics, there are too many variables and for many variables, data may be hard to get. There are complex issues such as water and air pollution, point and non-point source pollution, deforestation, climate change etc. In such a sector, the empirical studies would have to be very specific, more localised and rather narrow. For example, Ning Wang in her paper, Measuring Transaction Costs: An Incomplete Survey, has referred to Colby's (1990) study on transaction costs of water transfer from agriculture to other uses. In addition to the price of the water and the transfer cost, there are other costs such as attorney fees, engineering and hydrological studies and fees paid to state agencies which he classifies as transaction costs. This is a very narrow area (that of transfer of water from agriculture to other uses).
Another point in respect of empirical studies to measure transaction costs is that it varies with the level of development of a country. In developed countries, where the formal legal systems are mature, contracts negotiations and enforcement is sophisticated and data is more readily available, measuring transaction costs would be easier.
Finally, the type of activity/production/resource and its geographical spread would be important factors in measuring transaction costs. For instance, production in multiple locations would mean varied legal jurisdictions and hence higher transaction costs. Similarly, if the resource under management is a local fishery with well-defined users and clearly laid out rules for the use, it would be easier to measure the transaction costs, as opposed to a more diffuse resource such as river water.
Measuring transaction costs would also depend on the type of property management in place: in a privately managed production unit, the roles and responsibilities are clearly laid out, accountability is well defined and the objective function is clear (maximisation of profits or cost-minimisation) and hence measuring transaction costs would be easier. This is in contrast with public bureaucracies where roles and responsibilities are diffuse and the objective function is more general (usually promoting public interest) and transaction costs definition and measurement are accordingly more difficult. In common property management resources, particularly in smaller community and village settings, the boundary of the resource is clear, the terms and conditions of usage are known
to all and everyone has a stake in the resource (e.g. a local fishery or a community using the adjacent forest), measurement of transaction costs is again relatively easy.
It is clear from the above discussion that 'transaction costs' as a concept has come a long way since Ronald Coase first alluded to it in 1937. Thanks to the work of Williamson and subsequently of other economists such as Harold Demsetz, Armen Alchian, Douglass North, Elinor Ostrom, Avner Greif, Steven Cheung and Acemoglu, transaction costs provide us with a compelling alternative to the neoclassical narrative on the theory of the firm. The shift from 'exchange' to 'transaction' as a unit of analysis has undoubtedly enriched the theory of the firm. However, the empirical work on transaction costs is still a work in progress and has progressed much more slowly. In the discussion above, we have outlined some of the empirical studies undertaken and under what conditions, measurement of transaction costs is relatively easier.
'Transaction costs' is a central concept in New Institutional Economics and is closely related to the other components of NIE, namely, property rights, institutions, asymmetric information, contracts, collective action dilemmas, social capital and bounded rationality. We will turn to the other components of the NIE in the coming articles, particularly the empirical aspects of these components.
The writer is an IAS officer, working as Principal Resident Commissioner, Government of WB