Tuesday, December 23, 2003


Robert J. Gordon recently wrote “Five Puzzles in the Behavior of Productivity, Investment, and Innovation� (September, 2003). The paper examines the role of ICT in enhancing productivity growth in developed countries, especially total factor productivity (TFP). (Gordon thinks that there has been an ICT based increase in the rate of TFP growth in the United States, and wonders why the same effect has not been as evident in Europe.)

I wonder why Gordon focuses so heavily on increasing TFP growth rates. Is it not enough that ICT investments maintain TFP growth at previously established levels? It is rather well established that technological engines of TFP growth eventually run out of steam. Steam engines may have fueled the industrial revolution, but no one is investing in steam engines for their factories or locomotives any more. To keep productivity growth going, new technologies with further growth potential have to be found to substitute for those which no longer fuel growth adequately. ICT may be a key ingredient to fuel productivity growth in developed countries at this moment; the importance of ICT may be to sustain the rate of increasing TFP, not to increase that rate. The Information Revolution may be important in sustaining the technology fueled TFP growth by new means as the most recent previous technological engines run out of steam.

Lets look at the micro-level. A firm is making a decision on whether to invest or not to invest. There are generally various options, and generally each option has its technological implications. In developing countries, rates of return for investment are already high. Firms have often not made the transition to the 20th century, much less the 21st. They still can make high-yielding investments in cars and trucks, electrical machinery, and telephones, not to mention human, organizational, and social capital. Many such investments will have high rates of return, since the society is not yet saturated with these technologies.

Add to the opportunities from pre-existing technologies, new opportunities to invest in PCs and Internet connectivity. Of course the new technologies will provide opportunities that in some cases substitute for those provided by older technologies. One will probably not buy a typewriter when one has bought a PC and printer. But in general the new technologies will simply expand the range of investment opportunities.

There will still be a lot of investments in cars and trucks, electrical machinery and telephones, but there will also be firms that choose to invest in PCs. In some cases the firms will make the wrong investment decision – they will buy a PC when it would have been more profitable to use the money to buy a car. But in general, we assume that they will make the right decision, investing in more (rather than less) profitable technologies. Generally, the highest available return on investment adding ICT to the pre-existing technology mix should be no lower than that from the un-supplemented, pre-existing technologies alone, and may be higher.

But the gains in general may be expected to be marginal. If one can expect a 15 percent rate of return on investment in non-ICT technologies, how much will adding ICT to the mix increase the average rate of return? Probably not a lot. Some funds that would have been invested in lower-yielding technologies will be shifted to higher yielding ICT, but most investments will not change. Firms would have bought cars and electrical machinery had there not been ICT available, and will still buy those things with ICT available. Only a part of investment will be in ICT, and that part will have only somewhat higher rates of return on average than would their non-ICT best alternatives.

Of course, with higher returns on investment available, one can expect some resources that otherwise would have gone for consumption now going into investment, and this shift can result in an improvement in labor productivity, and ultimately in more rapid increase in per capita GDP. But this effect too is likely to be at the margin.

With millions upon millions of investors making such decisions, there may be very large numbers of decisions to invest in ICT rather than in a more traditional technology, and very large number of decisions to invest a little more in the firm. These decision for ICT innovation are thrown together with the larger numbers of decisions to invest in traditional technologies into the complex processes of the economy. What emerges, among other things, are probably higher rates of labor productivity and TFP growth – i.e. more rapid economic development. But only marginally more rapid rates of development!

And as I have argued in the past, the potential benefits from ICT will only be realized where other conditions are propitious. Adding ICT to the technology mix will not add to investment where the investors are facing social, political or other conditions that make them unlikely to invest in any technology.

The donor community seeks evidence that ICT will enhance development. Measuring the impact of a technology on economic development is very difficult. (Measuring the impact on social, political, or cultural development may be a whole lot harder!) Complexity theory has illuninated the unintuitive nature of emergent properties of small changes in the way decisions are made in complex systems. Thus complexity theory helps explain why it is hard to establish the causal linkage between the introduction of new technological opportunities and the rate of economic growth. It is perhaps even more difficult, recognizing poverty as a multidimensional problem, to demonstrate the impact of ICT on poverty reduction. The difficulties of its demonstration should not be taken as arguing against the reality of the effect.

No comments: