What's the difference between short-term and long-term?
[Fairly long post: scroll to the bottom for a one-paragraph summary]
I've been interested in this debate for many years. It takes several different forms, but the simplest statement is: stockmarket investors are too focused on short-term returns at the expense of the long-term investment that builds real economic capital.
I've been interested in this debate for many years. It takes several different forms, but the simplest statement is: stockmarket investors are too focused on short-term returns at the expense of the long-term investment that builds real economic capital.
I first heard this argument deployed against "asset strippers" such as Hanson, a UK conglomerate prominent in the 1980s. Since then it's been used to describe stockmarket traders, company shareholders in general, financial institutions of all kinds, and US and UK companies who supposedly focus greedily on quarterly earnings, unlike the more virtuous French, German and Japanese firms who are willing to take the long view.
The first problem with the argument is this: long-term returns are just a string of several short-term returns in a row. For example, £1,000 invested over 20 years to return £5,000 is equivalent to an annual return of 8.3%. But at any point during that 20 years, the asset is worth more than the original £1,000. In theory, it can be sold at any time for a (short-term) profit. And the fact that someone sells it for a one-year return of 8% does not stop the 20-year investment from happening.
Of course some markets are not liquid. If I invest £100,000 in a high-speed die stamping machine on a 10-year business plan, I probably won't be able to sell the machine itself for £108,000 a year later. But that's why we wrap those assets up as shares in public companies, which can be bought and sold on any timescale.
Pharmaceutical research is an excellent place to see this process in action. It takes five to ten years, or longer, to develop a new drug. Sometimes large firms will do it internally, in which case the progress of drugs through the pipeline is reflected in the value of the company and investors can buy and sell as they wish. More commonly, a biotech startup kicks off the process, in which case you can clearly see the increase in value as the process occurs.
First, they raise venture capital to finance the initial development - at a valuation of a few million dollars. If they succeed, they enter a research or marketing joint venture with a big company at a valuation of $10-30 million after the initial results are proved. Eventually, they will usually sell out to one of the multinational firms for anywhere from $100 million-$1 billion. Finally, the accrued benefit to Glaxo's shareholders, through dividends or capital value, might be $5 billion or more.
This is a long-term investment, but financed by a series of short-term bets by different investors. Even within the two to three-year turnaround of a single stage in this pipeline, investors can buy or sell their stake according to their estimate of the likely success of the product. There is, in fact, no conflict between short- and long-term investment horizons in terms of how they are financed.
In other words, the financial markets are precisely how short-term investors can finance long-term investments. Conversely, a long-term investor can participate in a series of short-term investments made by the management of the companies whose shares they hold.
But moving away from the finance markets, there might be something behind this argument. A company does have a genuine choice between taking risky long-term bets on major new products, or making smaller innovations which will complete within a few months or years. And sometimes they choose not to make those long-term investments. But this isn't just risk aversion. There are some real facts about the world which could lead firms, rationally, not to make long-term bets.
[Update: in the comments, codemonkey_uk points out the challenge of local maxima. It is certainly possible that a short-term improvement can lead to a 'dead end' because it optimises for something that is true now, but might not still be true in ten years. This is one of the disadvantages of incremental development. But as he also points out, it's effectively impossible to see ten years ahead - so, like evolution, we are better off accepting this cost and pursuing short-term improvements anyway.]
Think of the intelligent designers' argument about the human eye. Evolution couldn't have produced this, they say, because it is too complex to have occurred in a random mutation. The individual components alone could not have evolved, because a retina provides no advantage without a lens, a cornea or the delicate muscles of the iris; and none of them in turn are any good without the retina.
However, it turns out that the eye could have, and did, evolve incrementally. In just the same way, most of our advanced and innovative technologies - the Internet, the car, the steam engine, the mobile phone - come from a long series of small improvements rather than a revolutionary fifteen-year plan.
The more basic fact is that technology (and investment) exist within a social context. Any technology is useless without the right context to enable it; Facebook would be pointless without widespread Internet access and a social desire to communicate, just as the spinning jenny is now useless in a world of electrically-powered machines. Since the social context is changing all the time, the best possible technology to make people's lives better is changing too.
This also means that a new application of an existing technology can be as useful as a new technology itself. Would anyone really claim that flying cars would make a bigger improvement to our lives than the Internet? If so, why aren't more people buying helicopters?
The fact is that many people have a romantic attachment to high-profile new inventions. The invention of the aeroplane is more important, they say, than the telephone's gradual eighty-year evolution from post-office or corner-shop speciality to one-in-every-home to one-in-every-pocket. But on most imaginable metrics - frequency of use, size of market, or impact on daily life - that simply isn't true. New inventions are more macho, and more visible in the historical record, than evolutionary improvements, but they are simply less significant on average.
Of course there is a role for both, but if we had to pick one, increments are better than revolutions.
So in the financial markets, short-term and long-term investment are exactly the same. But in the real world of decisions on how to create products, a gamble on a specific long-term goal is usually less effective than a series of short-term results that build on each other and, in the long run, change the world.
Of course some markets are not liquid. If I invest £100,000 in a high-speed die stamping machine on a 10-year business plan, I probably won't be able to sell the machine itself for £108,000 a year later. But that's why we wrap those assets up as shares in public companies, which can be bought and sold on any timescale.
Pharmaceutical research is an excellent place to see this process in action. It takes five to ten years, or longer, to develop a new drug. Sometimes large firms will do it internally, in which case the progress of drugs through the pipeline is reflected in the value of the company and investors can buy and sell as they wish. More commonly, a biotech startup kicks off the process, in which case you can clearly see the increase in value as the process occurs.
First, they raise venture capital to finance the initial development - at a valuation of a few million dollars. If they succeed, they enter a research or marketing joint venture with a big company at a valuation of $10-30 million after the initial results are proved. Eventually, they will usually sell out to one of the multinational firms for anywhere from $100 million-$1 billion. Finally, the accrued benefit to Glaxo's shareholders, through dividends or capital value, might be $5 billion or more.
This is a long-term investment, but financed by a series of short-term bets by different investors. Even within the two to three-year turnaround of a single stage in this pipeline, investors can buy or sell their stake according to their estimate of the likely success of the product. There is, in fact, no conflict between short- and long-term investment horizons in terms of how they are financed.
In other words, the financial markets are precisely how short-term investors can finance long-term investments. Conversely, a long-term investor can participate in a series of short-term investments made by the management of the companies whose shares they hold.
But moving away from the finance markets, there might be something behind this argument. A company does have a genuine choice between taking risky long-term bets on major new products, or making smaller innovations which will complete within a few months or years. And sometimes they choose not to make those long-term investments. But this isn't just risk aversion. There are some real facts about the world which could lead firms, rationally, not to make long-term bets.
- Uncertainty. It is very hard to know what kind of products will be demanded in twenty years. The economy will change beyond all recognition. Could Apple have possibly known enough to start investing in the iPhone in 1989? Of course not. But could it invest in the Mac Classic, a more consumer-friendly and cheaper version of its then-mainstream product? Yes. And from there, the iMac, a brief diversion via the failed Newton, back on track with the Powerbook, the Macbook and finally the iPhone when they could see that people would actually use it. Most revolutionary new inventions fail.
- Incrementality. Perhaps it's actually better strategy to invest in a series of small improvements rather than trying to achieve one big one. Cars now are much better than cars thirty years ago, but not because BMW set out on a huge long-term investment programme to invent "the car of the future". Instead, they carried out a series of small improvements - each of which was very likely to be successful on its own terms, and their product kept getting better. (On the other hand, we don't have those flying cars yet. You could argue that if they had started on that in the 1980s, it might be done by now.)
- Compounding. Maybe the iPhone is five times better than whatever we had in the 80s. But does that mean Apple should have built it, even if you could predict that it was going to be successful? Perhaps not. If you can instead make your current products 10% better every year - and I'd argue that Apple has done exactly that - then in twenty years the outcome will be that your products are seven times better. Major new developments might look spectacular, but they are not the only way to make big changes in how we live.
[Update: in the comments, codemonkey_uk points out the challenge of local maxima. It is certainly possible that a short-term improvement can lead to a 'dead end' because it optimises for something that is true now, but might not still be true in ten years. This is one of the disadvantages of incremental development. But as he also points out, it's effectively impossible to see ten years ahead - so, like evolution, we are better off accepting this cost and pursuing short-term improvements anyway.]
Think of the intelligent designers' argument about the human eye. Evolution couldn't have produced this, they say, because it is too complex to have occurred in a random mutation. The individual components alone could not have evolved, because a retina provides no advantage without a lens, a cornea or the delicate muscles of the iris; and none of them in turn are any good without the retina.
However, it turns out that the eye could have, and did, evolve incrementally. In just the same way, most of our advanced and innovative technologies - the Internet, the car, the steam engine, the mobile phone - come from a long series of small improvements rather than a revolutionary fifteen-year plan.
The more basic fact is that technology (and investment) exist within a social context. Any technology is useless without the right context to enable it; Facebook would be pointless without widespread Internet access and a social desire to communicate, just as the spinning jenny is now useless in a world of electrically-powered machines. Since the social context is changing all the time, the best possible technology to make people's lives better is changing too.
This also means that a new application of an existing technology can be as useful as a new technology itself. Would anyone really claim that flying cars would make a bigger improvement to our lives than the Internet? If so, why aren't more people buying helicopters?
The fact is that many people have a romantic attachment to high-profile new inventions. The invention of the aeroplane is more important, they say, than the telephone's gradual eighty-year evolution from post-office or corner-shop speciality to one-in-every-home to one-in-every-pocket. But on most imaginable metrics - frequency of use, size of market, or impact on daily life - that simply isn't true. New inventions are more macho, and more visible in the historical record, than evolutionary improvements, but they are simply less significant on average.
Of course there is a role for both, but if we had to pick one, increments are better than revolutions.
So in the financial markets, short-term and long-term investment are exactly the same. But in the real world of decisions on how to create products, a gamble on a specific long-term goal is usually less effective than a series of short-term results that build on each other and, in the long run, change the world.
Comments
http://en.wikipedia.org/wiki/Hill_climbing
In all optimization problems, be it evolution, or financial planning, the further you can accuratly look ahead then the better the result.
The problem with long term planning is that (a) it becomes exponentially more expensive the further you look, and (b) in a configuration space as complex as financial markets, it is virtually impossible to look ahead with any accuracy at all.
Long term planning fails, not because short term planning is *better* but because we cannot see into the future.
Evolution doesn't give the best "solution" it gives the solution that is good enough. The human eye example is brilliant in showing that, because it is so riddled with "design" flaws, a situation that can only have come about as a result of evolution. For example, a well designed eye would have no blind spot.
I fully agree that local maxima are a critical issue in optimisation and decision-making, and am surprised to see that I haven't mentioned the phrase in any of my previous postings (though this one touches on the same point using different phrasing - search for the word 'local').
On a technicality, I would take issue with this comment: "Long term planning fails, not because short term planning is *better* but because we cannot see into the future." This failure of long-term planning is a clear demonstration that short-term planning is better - by default at least!
Thanks, and I will update the post to reflect your comment.
An owner of a car who only owns the cars for short terms is not going to invest large amounts of money in the up keep of the car, they will sell the car when the investment costs become to high. Now repeat this several times for the same car and soon it's worthless and the life expectancy is for shortened because of bad maintainence (investment).
The opposite happens for a long term car owner - the life expectancy is increased and the long term investment is lower as a new car is not required.
This never ever happens in the companies market does it!
Good example. My argument is that even if the first owner (A) doesn't plan to keep the car for a long time, it is still in their interest to invest in maintenance because they will get a higher resale value when they sell it to the next person (B).
But there are at least two reasons why this might not happen.
1. Asymmetric information. B might not easily be able to tell whether A has taken good care of the car. If so, A might as well not bother, as the extra money or time spent on maintenance will not be reflected in the sale price.
2. Hyperbolic discounting. It requires a certain amount of time and trouble, as well as money, to keep a car in good condition. While A is driving around through slush, rain and grit, she may not be thinking about the resale value, being focused instead on quickly getting home and out of the weather. This may result in a lower-than-optimal amount of maintenance or care in driving.
The second effect will always happen, no matter what; but the first will definitely be less likely if A knows they are going to own the car for a long period.
In the corporate case, this is a good argument for transparency and liquid markets. But those might not be enough to solve the problem - you make a valid point.
The second point, if anything, argues for short-term returns, because when the costs and the returns are more closely aligned, people will make better decisions.
The countries that planned long term (Japan, Germany) have done better than the countries that clear-cut their forests (Easter Islands).
I am sure lots of other examples exist where optimizing short term results results in a permanent "opportunity cost" and loss off greater long term return.
In cases where such markets are not available to send long-term signals "back from the future" as it were, it can certainly be possible for short-term decisions to lead to disaster.
It would be interesting to develop a balanced theory of how to mediate between short-term and long-term goals. In a predictable world where good information is available and markets are complete, the standard theory of risk-weighted discounting provides an answer.
In the real world, where information is asymmetric, transaction costs are high, cognitive biases are rife and the future is highly uncertain, we need to develop better insights.