Why GDP in a service economy cannot grow, or productivity paradox explained.

In Economics, Productivity Paradox (why introduction of computers resulted in no productivity increase) still has no widely accepted explanation. Naturally, I must post my own here. I think this longish essay will be worth the read.

First, imagine an economy where over 50% of people are employed in food production; most of others are creating tools and houses; and "service sector" is an expression that nobody has heard. If you think this has nothing in common with the economies of today's developed countries, you are absolutely right. Yet this 19th century economy is exactly what people had to work with when most of the traditional economic theory was founded. In this economy, it is easy to calculate things. If total amount of money is M and total product is P, then worker's salary is typically (P/M)*V, where V is how much product that particular worker is responsible for. If you then add up all the salaries in the country, you get back something that is directly related to the total amount of physical goods produced. This is the GDP, measured in dollars, and it captures the country's output very well. Let's call this exhibit Economy #1.

Now fast forward to the day when technology advanced to the point where all fertile land can be managed by 2% of the population; all necessary clothing  can be made by another 2%; all required tools and houses can be built by another 6%; what about the other 90% who have nothing to do?

That's when the service economy is born, where most work is done not because it's needed, but simply in order to justify getting your fair share of food and other stuff. Most jobs pay based on how long they take to complete and not on their "usefulness", and it is therefore possible for a completely useless service (such as real estate brokerage) to be priced higher than somewhat useful ones (like lawn mowing and hair cutting). Producers of real goods have no choice but to go along with the program because otherwise they would not be able to sell anything to anyone (and would get lynched by an angry mob, to boot). In this economy, most of the product is intangible; and the total physical product P is a negligibly small part of the economy. In fact, the only thing produced by this economy is the time that people spend servicing each other. If total population is Y and total revolving money supply is M, then the average salary in this economy is going to be simply M/Y - because it's the only way to distribute food and shelter evenly. Notice how drastically different that is from our Economy #1. Even though some people (*cough* lawyers) will be better than others at marketing their time, the average is always going to be M/Y, and if you measure the GDP traditionally, by adding people's salaries, then total GDP is now completely decoupled from the actual physical product, and depends only on the money supply. In this economy it is practically impossible for the real output to grow, because the only significant product is people's time, and that is a fairly inflexible thing. And while you can "improve" GDP by creating more people or making them work more hours, it is impossible to change the productivity - because it will always be proportional to M/Y, no matter what happens to the total amount of tangible goods! And since we measure productivity in money, we will always get the same number back.

Productivity paradox is explained trivially once you understand that. If you add computers to an office that employs 100 people and it now takes 33 people to do the same job, it does not triple the salary of the original 100 workers, because remember, they are not producing anything, they are just trying to sell their time to justify getting the appropriate amount of food and physical goods. If their jobs suddenly paid better, more and more people would apply for them and drive the salary back to average. So what must happen then is 77 workers have to be laid off and go on to become personal trainers, dog walkers and so forth. Measuring the productivity of the original 100 workers will give you exactly the same number before and after - because we measure it in money. The elusive productivity growth is actually represented by more services becoming available - but we have no tools to measure that.

You might ask, why is reported GDP changing, if productivity must remain constant?
Since money supply M is always changing, governments adjust productivity by inflation. However, they don't use change in M as their inflation number (and it's impossible to measure it anyway). Instead, a basket of goods is used to approximately measure the change in money supply, and of course it's a very crude proxy. So all changes in GDP that we are observing are actually just mismeasurement of inflation.

---

As a bonus, we can now explain another paradox. You may have heard that pumping money into economy increases GDP, but in traditional economic theory that shouldn't work. In Economy #1, so many people are producing food and tools because they are actually hard to produce in low-tech society, and there isn't enough for everyone. So pumping money into economy (i.e. giving people more money) instantly increases demand for food and tools, and that drives up inflation by the same amount as the money supply, so total GDP (adjusted by inflation) does not budge.

However, if you add money into economy #2, people have no incentive to spend it on food or cloth, since people are already eating more than they need, and have more clothes than they can wear. So instead they dump their new money into things like bigger houses, they bid up education and health services, they buy stocks. Many of those things will not be (fully) captured by the traditional basket-of-good inflation measures, and therefore the GDP number is going to swell up instead. But it's all just an illusion - simply a mismeasurement of inflation. And that is exactly what happened in the USA in 1995-2006 and it explains how it's possible for the standard of living do stagnate during a period of GDP "growth".

Happy Valentine's Day!

Sign of the times, from New York Times:

[...] on Yahoo, searches for “cheap engagement rings” are “off the charts” compared with a year ago, according to Vera Chan, a trend analyst for the company.Other searches that are up over last year include “cheap lingerie,” “free Valentine’s Day cards” and “homemade Valentine’s Day gifts.”

Cheap engagement rings sounds almost like an oxymoron. But I liked that part:

Personal jewelry is being replaced by personal poems.

Maybe that's good news. So on that happy note, happy Valentine's Day to everyone!

Projecting the severity of the recession from credit market spreads.

When I was reading about the Great Depression in US I was struck by the following observation: the GDP decline of 28% peak to trough, which took somewhere between 2 and 3 years, correlates very well to the total number of failed companies (30%) and the peak unemployment (25%). After thinking about this for a while, I believe this is not a mere coincidence. In a severe, quick economic contraction, it would make sense that the number of working people directly relates to the GDP, since productivity does not have time to change much. It is also reasonable that the percentage of failed mid-size companies reduces GDP by the same amount, although this requires that the additional contraction among the surviving companies is offset by increased government activity and small businesses started by newly unemployed.

So while the number of failed companies is not a precise predictor of the depth of a depression, it should give the right ballpark number. Since we can deduce company failure rates by their bond spreads, one can therefore estimate market-implied severity of a recession.

From Financial Times:

US investment-grade corporate bond prices, for example, imply a cumulative default rate of 36 per cent over five years, assuming a typical recovery of 40 cents in the dollar, according to analysts at Morgan Stanley. This is more than 7.5 times higher than the worst default rate in any previous five-year period.

5 year default rate is not horribly useful, since a lot can change in 5 years. So let's convert it to a more relevant metric of roughly 8-9% of defaults per year. I think we can safely assume that the rate at which new mid-size companies are created goes down to roughly 1% from the typical 2-4% in 'normal' years. This means that within the next 2 years (typical time-frame for the worst part of recession) we are looking at the number of investment grade companies in US to shrink by roughly 15%.

So the market is pretty pessimistic and investors seem to project a recession with GDP decline in the ballpark of 10-20%, which puts it squarely in the depression camp.