Saturday, July 02, 2011

Swedish monetary policy: a success

Nominal GDP targeting seems clever

Sweden is one of the EU countries that didn't enter the eurozone and it's arguably the most healthy country when it comes to the recovery from the recent recession. Sweden would be led by the socialists for decades which didn't destroy the Scandinavia country for many reasons (although their crisis in the early 1990s wasn't too nice). The private sector was kept very strong and the Swedes are just hardened enough people.

However, in the recent years, Sweden has been led by the center-right "moderate" party and there are other reasons why Sweden's economy got even better.

Patria Finance is a top Czech trading company - its ex-president was sitting next to me at school for 8 years :-) - and I learned about the Swedish monetary policy that's been very successful. Among other good signs, the recent quarters displayed a 7% real annual GDP growth and the unemployment is close to 7%, too.

And make no doubts about it, 7 is a pretty lucky number in both of these contexts.

The previous link leads to a Czech page and you may find it annoying. So let me list a few sources which are both more original and in English:
Neil Irwin at Washington Post

David Beckworth at his blog

Matt Yglesias at Think Progress
David Beckworth is an assistant professor of economics. Otherwise you can see that I chose pretty left-wing sources. But they're right about the essential points, I think.

Their central bank wasn't afraid to make its balance sheet up to 25% of the annual Swedish GDP - it is just 15% in the U.S. During the recession, it was effectively pumping more money to the system than others. But what's the most interesting quantitative insight about their behavior is that they're effectively targeting the nominal GDP.

In essence, the bank prescribes that the GDP grows by 5 percent every year. If the observed growth is slower or faster than the target, they loosen or tighten their monetary policy in the following quarters, to get back to the trend curve. Just to be sure, the 5 percent growth is the nominal growth - as expressed in the Swedish crowns. Targeting real growth isn't possible. If it were possible, many countries would target real growth above 100 percent per year. :-)

Research and debate about optimum rules for the central banks is just getting started

Just fourty years ago, people were trying to peg the value of their money to gold which is just another yellow metal. For economists, 40 years may be too short a period of time so there hasn't been a genuine progress in the thinking about those matters. Many people still think about the monetary policies as an adjustment or distortion of some kind of a gold standard that they still consider to be their zeroth approximation. They don't really try to get a better zeroth approximation yet.

However, the invisible hand of the free market may show its power regardless of the behavior of the currencies that are used for most payments. In particular, capitalism may produce growth in the environment with a stable currency, inflating currency, and even deflating currency.

One may talk about the "principle of relativity". The inflation is analogous to a velocity in mechanics and it doesn't matter. If you increase the inflation rate and all other nominal rates - especially interest rates - by the same amount which is constant in time, you get an equivalent arrangement. The real rates - various growth and interest rates minus the inflation rate - will adjust themselves to be equal in both situations. All inertial systems will experience the same laws of physics - and economics.

There is one subtle exception: it may be hard to make nominal interest rates negative because people are apparently always able to guarantee at least zero rates. Just move all the cash into your pillow. However, Sweden has actually shown that this exception isn't quite necessary because at some moment, they effectively imposed negative nominal rates for all large deposits that were sitting in banks, doing their best not to help the economy that needed some money supply.

So unless you get to the situation in which billionaires store their billions under the pillows, negative rates are actually possible and the "principle of financial relativity" holds completely.

Just like in Newton's mechanics, accelerations are real and independent of the inertial frame. So the increase of the money supply e.g. by lowering the benchmark interest rates will eventually cause the inflation to rise - or deflation to disappear - and people will be more willing to lend their money, too.

Now, if you overtake the Parliament and other institutions, you may impose pretty much any system with any monetary policies. However, that doesn't mean that all of them are equally good. What does it mean for a policy to be good? You want to optimize the efficiency of the markets, to unlock the human potential etc. It may be hard to define these notions quantitatively but one may get close to it. Very soon, I will outline some more detailed conditions that a good monetary system should satisfy.

The purpose of money is to be able to delay the consumption or, on the other side of the equation, to speed up the investment that will be repaid in the future. Those two processes - spending before you earn the money; and spending after you earn the money - should be kind of balanced. More precisely, the imbalance between them should behave "nicely" and it should certainly suffer from no unstable, runaway behavior signalling a lethal instability.

For everyone to be optimally able to do planning about the future, it's ideal if the currency used to borrow - or lend - is "stable". More precisely, we don't need the currency to be stable even though up to a few percent, it is stable. We need the value of the currency to be predictable.

Off-topic: Condi Rice opened a Ronald Reagan Street in Prague. That's something that they have in many places of Poland - but a street or another thing named after Reagan is something they can't afford in any other country outside the U.S. and not even in Massachusetts or Vermont, among others haha.

But the amount of money depends on the units and there are no "God-given" units. As indicated in the financial "principle of relativity", the overall value of the money may evolve in any way - the unit may be time-dependent, at least exponentially increasing or decreasing - and all physical effects of such a change may be compensated by the correspondingly higher interest rates etc. A good currency system should satisfy at least the folowing conditions:
  • the value of the money should be sufficiently inclusive - or as inclusive as possible - so that the maximum number of the market players may avoid a maximum fraction of the speculations about the future evolution of price ratios that have almost nothing to do with their primary transactions
  • the future value of the money should be linked to something physical that can be imagined regardless of the choice of the units
  • the future evolution of the value of the money should be sufficiently predictable
  • the predictions shouldn't include instabilities or predictably accelerating processes
  • for practical purposes, you don't want the inflation to exceed several percent because it leads to the inconvenient need to change the prices in the shops too often and/or to print new banknotes with additional zeroes
It's much better if the central bank policies are almost automatic and don't depend on arbitrary idiosyncratic opinions of a few people. This statement has several reasons: first of all, you don't want to give a few people too much power. Second of all, such bets on a person's psychology bring lots of unnecessary volatility to the markets. Third of all, the people doing such decisions might try to benefit out of the motion of prices - or their friends could benefit. Fourth of all, individual decisions are likely to suffer from systematic errors - they're very likely to make the policies "systematicaly more dovish than appropriate", if I use the most common mistaken approach of the central banks which affects a vast majority of the "happy times".

(The Swedish central bank not only allowed the money supply to grow quickly during recession; it also used a correspondingly tight and principled set of policies during the previous happy times - something that gave them lots of room for maneuvring.)

Targeting: what to target

Fine. So recall that we're designing a clever monetary system that makes people maximally happy and allows them to use their potential. We want the policies to be automatic enough in order to limit the power of individuals. But we still have infinitely many possible strategies, formulae, and algorithms - and all their conceivable mixtures and combinations.

When we're targeting something, we want to make a certain "nominal" quantity evolve according to a simple pre-determined exponential formula.

Most modern central banks are targeting inflation. The general price level should fluctuate around 2 percent. These days, the nominal interest rates are typically lower than that so the real ones are negative but that's another issue.

Inflation targeting is a natural "generalization" of the gold standard. Instead of making the price of gold constant - which is silly and arbitrary - the central banks want to make the price of an inclusive basket that contains lots of things constant - more precisely, increasing at a fixed rate such as 2%.

There are lots of technical questions how the inflation should be calculated and what should be included in the basket. Should it include volatile things such as food and fuels? I think it should even though it makes the balancing harder. Should it include stocks or houses or land? Maybe.

But it's plausible that there exist completely different philosophies that could actually lead to better financial environments. The GDP targeting is a possible example.

In the gold standard and inflation targeting, we're looking at the price of some "fixed things". However, during recession, the price doesn't usually change much because despite the dropping demand, it doesn't become much easier to reduce the price. Despite the small impact on the price level, the system badly needs much more money supply which has become too low. And on the contrary: during happy times, there may exist bubbles despite a low inflation which reflect the fact that there's too much money supply in the system. Those bubbles are usually allowed to grow because the price of the bubbling assets isn't included in the inflation basket, so the sick behavior isn't visible to the bankers who just do inflation targeting.

The GDP targeting satisfies pretty much all the conditions we expect from a good system. It really determines a sensibly constant money supply - it sensibly balances the interests of the people who have delayed their consumption and those who sped it up. The GDP targeting paradigm deals with some very inclusive numbers - the total GDP expresses everything that the economic subjects have sold during a period of time - e.g. a quarter or a year. So it is not linked to some arbitrary things such as gold. Moreover, the GDP kind of automatically creates a sensible "basket". It's a basket that includes "really everything": all kinds of income - which is taxable after you subtract the expenses - is a part of the GDP.

If a central bank prescribes a 5% nominal GDP growth rate - and the Riksbank has effectively done something equivalent - it means that you may rely on the value of the money. The value is predictable in some units you may imagine. If you're promised to be paid X Swedish crowns in the year 2020 - assuming that the country will avoid the euro membership (or assuming that the ECB will adopt similar policies by 2020, unless the eurozone decays completely) - you know that it means that you will get X.C% part of the Swedish GDP where both X and C are known in advance.

The expected average is that those 5% will include a 3% real growth and a 2% inflation. But of course, you can't strictly assume both of these numbers separately. When it comes to the future value of your fiat currency, you're allowed to impose at most one condition - and ideally exactly one condition, of course. More than one condition means an overdetermined system i.e. mutually incompatible constraints on the value of the money.

If the Swedish crown is - effectively - linked to the nominal GDP, you know how big a part of the economy you will have to represent for you to be able to pay a certain amount in 2020. This is a pretty well-defined task you may be able to predict. In the same way, if you will be paid X Swedish crowns in 2020, you know what part of the GDP you will get. Those considerations show that the planning is easier both for those who promise to pay some money as well as those who are promised to be paid some money in the future.

What you don't quite know is how much money you will have if you will save them now - because this is determined by the real interest rates. The real interest rates - and in fact, even the nominal interest rates - are strictly speaking unknown. Again, we're assuming that the nominal GDP growth is predetermined so we can't make additional assumptions because all other quantities and rates will be determined by the free market. Your guesses what they should be are irrelevant; you can't prescribe another condition constraining the value of the money.

But you may still qualitatively see what will happen with your savings under different scenarios. If the real GDP growth will be large, inflation will be low because the nominal GDP growth - the sum of the real GDP growth and inflation - is fixed. However, in that regime, the benchmark (and other) interest rates will probably also be low. And vice versa. If the economy is doing badly, the real GDP growth is low, the inflation will be pushed to high values, but so will the interest rates.

At the end, the real interest rates won't really depend on the real GDP growth much. After all, that's true regardless of what you're targeting.

GDP targeting vs money linked to stocks

In the past, I was promoting "stock market index targeting" - namely that a stock market index should be kept constant. This is similar to the GDP targeting because both of them involve the value of some "large and fast capital".

But when I compare the stock market index targeting with the GDP targeting, I think that the latter is better. The value of the stocks should be inherently volatile because the price of stocks depends on the profit and therefore the profit margin etc. Those things inevitably depend on the state of the economy. And the investment into stocks - even all stocks - must inevitably be risky. All owners of companies must be "bolder people than the rest" so you can't link the savings of average people to the assets of the bold owners of companies.

GDP targeting is even more inclusive and the GDP really expresses all the money that is being moved from one place to another in recent months.

What I want to emphasize is that if you have fixed and good enough rules - such as GDP targeting - you may show that the system won't collapse even under extreme circumstances and it will still protect some long-term predictable value of the money. Because the value of the money is ultimately linked to something - the GDP in this case - you know that short-term oscillations are of temporary nature. Because you know that they're of temporary nature, you may allow the system to do really extreme things with the interest rates etc. And indeed, such an optimum system may often behave in more extreme ways than the systems we're used to - and it might be the right thing.

For example, as I mentioned, you may allow the interest rates on large lazy savings to go negative which has occurred in Sweden.

In monetary systems without a clear theory and a clear philosophy - and the Federal Reserve System is an example - the loosening of the policies always seems extremely risky because you don't know whether it will end, where it will end, and what will be left out of the U.S. dollar.

And because of that, the systematic errors may go in both directions. During the happy times, it's normal for the policies to be too dovish which often leads to various bubbles. But in the hard times, there exists a psychological limit and the monetary policies are never loosened too much. They were loosened much more in Sweden than they were loosened in the U.S. - and it was still safe because the money is still linked to the nominal GDP according to a simple formula that exponentially depends on time.


GDP targeting seems to be the most natural prescription for a modern monetary policy that I've encountered, at least among the "simple ones". It guarantees a predictable value of the money but it still automatically regulates the money supply in the economy - more aggressively than what the systems based on inflation targeting do. There might exist better rules - hybrid rules etc. - but the macroeconomists and the real-world central bankers should begin some serious research into these matters.

The optimum currency systems may significantly deviate from the "traditions" because the traditional policies could have been vastly suboptimal. Despite the fact that such a modern system may be changing interest rates and money supply much more dramatically than the traditional systems - in both directions (and, which is related, the balance sheets of the central banks may be much larger, de facto unlimited) - its internal structure may be enough to prove that it cannot lead to vicious circles and that it allows a very comfortable financial planning for the future.

As the vice-governor of the Czech National Bank told me during a lunch a year ago or so, the research into these matters is just getting started. I hope that he is personally thinking about the consequences of GDP targeting, too.

Appendix: targeting GDP per capita

One may design a variation of the GDP targeting - targeting of GDP per capita. Which of them is better? Now, lots of irrational arguments could be used. People could decide that one of them is better and they like it, and because the other proposals are different, they have to be wrong.

Well, that's not a way to decide. The only rational way is to consider a general enough situation in which the two policies differ - and decide which of them is more well-behaved.

In particular, we have to consider what happens if the population of Sweden changes. This hasn't been a big issue in reality because the population growth in Sweden is not too large and is nearly fixed. To make the thinking more imaginable, change the population a lot. Imagine that all the Finns move from the eurozone to Sweden.

Clearly, the real GDP changes a lot - almost doubles. So GDP targeting would mean that one Swedish crown would have to double. High interest rates would have to be imposed to hit the target. A lot of discontinuity and havoc would follow.

It's therefore probably better to target GDP per capita.

For Finnland and Norway, it could be OK - assuming that one counts all the GDP of the people who actually primarily use the Swedish currency. However, imagine that India - with a very different GDP per capita - wants to join the Swedish currency system.

Well, then even the GDP per capita targeting will lead to huge discontinuities. For such situations, it would probably be better to be prescribe a more sophisticated formula that prescribes a new GDP whenever there is a discontinuity in the population - so that the total real GDP of the "same people" behaves continuously. I could tell you quantatively what it means for the targets.

Appendix: gold price targeting

You could also think about adjusting the interest rates so that the gold price stays within a predetermined interval. Except that the changes of the interest rates would have to be extreme in many cases (plus or minus a dozen of percent of interest rates) - the gold price easily quadruples or goes to one quarter in a decade - and these modifications of the interest rates wouldn't really be useful for regulating the economy and its cycles. They would only be useful to regulate a yellow metal.


  1. I would definitely choose a stable currency with 0% inflation target (or according to population count change). But one main problem is IMHO that the central banks are giving money to the private banks. Sometimes as you mentioned with negative interest rates. Central banks should give money (lend) only directly to the government. Why are prefered the private banks instead of the public?

    If there were a GDP target I would suggest something from 0 to 1%. But GDP targeting is IMO not accurate in tax avoiding economies (black).

  2. A free society would not have a currency that is issued by a central bank, which is given a monopoly over its creation. It would not have some political appointees engage in top down central planning that picks winners and losers. In a free society the markets would determine what is used as money and would control the money supply. Plans would be bottoms-up rather than top-down.

    The debate about this is a very old one. The easiest way to get an overview is through the two great videos done by John Papola and Russ Roberts. The first has already had more than 2 million views, which is impressive given the very technical issues being presented and the fact that the people most interested in those issues do not like the rap format which covers those issues. The second video, which was released not very long ago, is closing in on
    one million views.

    >"Fear the Boom and Bust" a Hayek vs. Keynes Rap Anthem

    Fight of the Century: Keynes vs. Hayek Round Two

    I think that the biggest point that you can get out of this debate is the idea of a fatal conceit. As Hayek pointed out in his Nobel acceptance speech, If man is not to do more harm than good in his efforts to improve the social order, he will have to learn that in this, as in all other fields where essential complexity of an organized kind prevails, he cannot acquire the full knowledge which would make mastery of the events possible. He will therefore have to use what knowledge he can achieve, not to shape the results as the craftsman shapes his handiwork, but rather to cultivate a growth by providing the appropriate environment, in the manner in which the gardener does this for his plants. There is danger in the exuberant feeling of ever growing power which the advance of the physical sciences has engendered and which tempts man to try, "dizzy with success", to use a characteristic phrase of early communism, to subject not only our natural but also our human environment to the control of a human will. The recognition of the insuperable limits to his knowledge ought indeed to teach the student of society a lesson of humility which should guard him against becoming an accomplice in men's fatal striving to control society - a striving which makes him not only a tyrant over his fellows, but which may well make him the destroyer of a civilization which no brain has designed but which has grown from the free efforts of millions of individuals.

  3. Kamil, you can't just impose all these numbers, they're determined by the free markets, with exactly one exception which is the overall normalization of the mone - and its evolution in time.

    Requiring zero inflation or zero GDP growth target is preposterous because you would surely need systematically negative interest rates to achieve it and if people knew that the interest rates in banks would be negative, they would surely store the money in their safeboxes, including the billionaires.

    With a totally electronic currency that can't be stored in the pillow, it could be different, but as long as there are physical banknotes without an IP address and a detailed info on the internet, it is impossible to make a policy that implies systematically negatively interest rates.

  4. Kamil, you can't just impose all these numbers, they're determined by the free markets, with exactly one exception which is the overall normalization of the mone - and its evolution in time.

    But there can't be a free market when a central bank is given a monopoly over money creation and it is free to engage in activities that distort markets.

  5. Quite on the contrary. A modern economy where the profit margin is often very small - and has to be small - and where this profit margin derives from the difference between accurately calculated prices that have to be sufficiently stable and predictable, it's important to have a stable enough measure of the value, and an institution that has a de facto unlimited right to remove or add the money from/to the system is *necessary* for such a modern market economy to work.

  6. It is not a problem to have negative interest rates if you lend only to the state. If ppl save money that is good for them, if they can't get money from the central bank (by private bank). Saving money causes deflation, which is a positive feedback for negative interest rates. The rates can be adjusted then back to 0%.

    In my model private banks have money only directly from citizens or companies, businesses.