Why Polls are a Danger to Democracy

poll

Does this poll reflect the true preferences of the Dutch population?

It is 13 March 2017, and the intermediate polls of the Dutch election suggest that it is going to be a close call between the Dutch Liberal Party (VVD) and the populist Party for Freedom (PVV). Who is going to be the biggest party of the Netherlands? Both parties are currently leading in the polls, and the winner will lead the formation of a new government, and most likely provide the new prime minister. So the stakes are high.

Now suppose you don’t agree with the VVD. You prefer a more progressive party (such as GroenLinks or D66). However, one thing is for sure: you don’t want Geert Wilders (PVV) to win, let alone become the prime minister. What do you do, knowing that it is going to be a close call between the VVD and the PVV? Do you vote for the party that truly reflects your preferences (GroenLinks or D66 in this example), or do you vote for the Dutch Liberal Party, knowing that you prefer the Liberals to the Populists?

The last option (to vote ‘strategically’, or not on your most preferred candidate) is handed to you only because of the information you derived from the intermediate polls. It is only because you know what the result of the election will be – given that people will vote as indicated in the poll – that you can adjust your vote to it.

But this raises a question: do we want people to have the opportunity to vote strategically? Or differently: should we allow for intermediate polls?

Democracy as a reflection of voters true preferences

This basically comes down to another question: what do we want the election results to be a reflection of? Do we want the results to be a reflection of the true preferences of the members of a population (meaning: if 20% of the people most prefer the VVD, then the VVD will get 20% of the votes, etc)? Or do we want the results to be a reflection of both true preferences and insincere preferences (i.e., 30% of the people vote for the VVD, even though only 20% has the VVD as their preferred option, with the other 10% preferring the VVD to the PVV).

I think no reasonable person would say that we hold elections to elicit insincere preferences. After all: what is the point of a democratic election in case we want to end up with election results that don’t reflect the population’s true preferences? This might even contradict the notion of democracy itself. So we want to elicit the true preferences of voters through an election.

However, by publishing intermediate polls, we entice people to cast votes that don’t reflect their true preferences (i.e., voting for the most preferred candidate), as shown by the example above. Hence the election result will not be an accurate representation of people’s preferences, hence not an optimal form of democracy.

Bandwagon effect

But there is another argument to be made against polls: the bandwagon effect. The bandwagon effect implies that the rate of uptake of beliefs increase with the number of people already holding the belief. In practice this means that people, seeing a poll in which party x has more seats than party y, become more inclined to vote for party x than y, ceterus paribus. The result is that x gets more votes than it would have gotten in case the polls wouldn’t have been published, and another party – possibly y – less. Still: we end up a with election results that are different from what they would have been in case we didn’t publish the polls.

But you can also approach the issue from another angle by simply asking: what is the added value of polls? What do polls contribute to society? ‘Information,’ one could say: ‘information about the voting behaviour of the population.’ But what can we use this information for? We cannot use it to change the voting behaviour of others. So for that purpose it’s useless. It can only be used to change one’s own voting behaviour.

Banning polls

So it seems that the only contribution of polling is that it allows for changing your vote in light of the voting behaviour of others? Well, we have just established that this is not a good thing in a democracy. Therefore polling has no added value to society.

Assuming that all of the above is true, why then allow polls? Banning polls is not some far-fetched idea. It already happens in many countries, including France (on the day before the election) and Italy (15 days before the election).

And given the negative effect of polling on the democratic process, this might not be such a bad idea.

Why Central Banks Are Reaching a Dead End

Central banks have the duty to keep inflation below but near a certain level (2% in the case of the U.S. and the Eurozone). They use monetary policy for doing so. In case inflation gets too low, they decrease the interest rate, thereby spurring lending, hence inflation. In case inflation gets too high, they do the reverse. This policy is believed to be effective, and to have contributed to normalization of the inflation-rate in the past.

All well and good so far. But as anyone knows by know: there is something odd going on. We hit the 0% interest rate level in the U.S., and have even reached -0.4% in the Eurozone. This is a logical consequence of decreasing the interest rate more in times of accommodative policy than decreasing it in times of tightening: a downward trend, hitting 0%, and rates below 0%, at some point in time.

The down-trend in the interest rate

The year 2025

In principle, central banks could continue this downward trend in the interest rate indefinitely. In practice, most likely not. Lets take a look into the future, and see what would happen if we would continue this downward trend.

Suppose it is the year 2025, and, for whatever reason, we have (the risk of) deflation in the Eurozone. The ECB – having increased the interest rate from the level of -0.4% in 2017 – decreases the interest rate gradually from (let’s say) 1.5% to -1.5% (a decrease of 3%, which is less than the decrease after the crisis in 2007, hence not unrealistic). So now we have an interest of -1.5%.

Given such a low interest rate, either of two things are likely to happen, that we might want to avoid at all costs: the net income of retail banks gets literally nullified, or retail banks have to charge customers for keeping their money.

As shown in quantitative research, a decrease of 1 percentage point in the 3-month yield (which will occur in case of a decrease in the deposit rate to -1.5%) leads to a decrease in the net interest margin of banks (in advanced economies) of 60 basis points. What does this mean in case of our rate decrease to -1.5%?

Take ING (the largest Dutch retail bank) as an example. ING has a net interest margin of around 1.5% over 2016, resulting in a net interest income of around 13.2 billion euros, and a net income of 4.651 billion euros.

Inflation in the U.S.

A rate decrease to -1.5% will cause ING’s net interest margin to decrease to less than 0.9%, implying its net interest income will be 40% lower than in 2016, decreasing by +- 5.2 billion euros. Hence ING’s entire profit will obliterated. Assuming that the lion-share of cost-cutting has been done already in response to the crisis of 2007, this leaves little room to cut costs. And anyway: if the down-trend in the interest rate will continue, at some point the net interest income will become so low than no cost-cutting measures can make up for that.

Hence, decreasing the interest rate beyond some point will seriously put to risk the existence of retail banking in the Eurozone. Assuming that Draghi doesn’t want that to happen, the interest rate can never get lower then a certain level (-1.4% in our example – the real number might of course be different, but this doesn’t affect the principle at stake here).

Retail banks could of course adjust their business model, and charge people for holding their money. Suppose that, in case the interest rate gets below a certain point, banks will do so. How much can they charge people before people will withdraw their money and start putting it into safes? Let’s say that banks, due to lower net interest incomes resulting from the lower interest rate, charge customers 0.5% for holding their savings. If you own 200.000 euros, this comes down to 1000 euros a year. I can imagine this might already be sufficient for many people to start putting their money into a safe instead of their bank, hence causing significant draw-downs, and endangering the banking system. Again: the real interest rate at which this event might occur might differ from this example, but the principle still holds.

Assuming Draghi doesn’t want this to happen either, we can never lower the interest rate below a certain level.

What then?

But, assuming that we can never lower the interest rate below a certain level (which might be not too far in the future as showed by the reasoning above), what then can we do to deal with low inflation in the future? Lowering interest rates is no option any more.

One possibility would be fiscal policy. We could push the responsibility for maintaining a reasonable level of inflation from central banks to governments. But this might very well imply that governments have to borrow significant amounts of money to kick-start inflation. Judging by the amount of money pushed into the market by the ECB to increase inflation, this lending will lead to exorbitant increases in public debt, hence interest payments.

Central banks could of course buy other assets, such as equity, instead of bonds, when the interest rate has reached its lowest level. The problems from doing so are obvious: it might cause a significant disconnect between share prices and real business activity, resulting in bubbles. Furthermore, it might not be very effective in increasing inflation, because we don’t know how much of the money put into stocks will eventually reach the real economy.

Another option would be helicopter money. But what if the helicopter stops dropping money? Then we are left with higher prices, but not more money to pay for them, laying the foundation for another period of deflation.

Accepting deflation

We could also just accept deflation. We have been trying to fight it for over 30 years, like our lives depend on it. But do they? How problematic would it be for prices to decrease for an extended period of time? Would people really stop buying stuff, because it will become cheaper in the future? That seems unrealistic. People still need to buy food, clothing, and still need a roof above their head. So deflation might not be such a big issue after all – unless it becomes extreme, of course.

It is choosing the least worst option among a bunch of bad options. Hence we should not rule out any response to low inflation because its effects are undesired: it might still be the best option.

One thing seems crystal clear: we have to think about future responses to low inflation, because the path we have currently taken seems to be near reaching its end point.