Rational Partisan Theory, elections, fiscal consolidation and exchange rate determination

When I was in university more than 20 years ago I had one other major interest other than monetary economics and that was public choice theory and I was particularly interested in how to understand macro economic issues through the use of public choice theory.

I particularly remember writing a paper on Political Business Cycle (PBC) theory and at some point I was even considering writing my master thesis on this topic. I instead ended up writing about Austrian Business Cycle theory – partly because I had grown somewhat disillusioned with the theoretical and particularly the empirical aspects of PBC (paradoxically enough writing my Master thesis had a similar impact in terms of leaving me utterly disillusioned with Austrian school macroeconomics).

At the core of my problems with the state of Political Business Cycle theory (at the time) was that even though I essentially was attracted to the traditional PBC model (as originally formulated by William Nordhaus in his is 1975 article “The Political Business Cycle” in Review of Economic Studies) I found that even though I liked to think of policy makers as somebody who try to maximize their power, influence and votes through distorting the macroeconomy I had (huge) problems with the macroeconomic framework – 1970s “Keynesian ” macro models – Nordhaus and other early PBC pioneers used.    

Hence, I might have liked the general political-economic ideas in the early PBC models, but I didn’t think that the macroeconomics of these models made much sense. At the core of this problem is of course that if it is so obvious that governments will ease fiscal (and monetary) policy ahead of elections to spur growth why is it that the agents in the economy (employers, investors, consumers and labour unions) does not realize this in advance? Anybody who had studied rational expectation theory of any kind would find it hard to believe that one systematically would be able to cheat labour unions into accepting lower real wages ahead of elections.

Said in another way if you introduce forward-looking agents in your models the Nordhaus style PBC models simply will not work. This of course in the late 1980s and early 1990s led to the development of models of the political business cycle that took into account the forward-looking behavior of economic agents. Most famously Alberto Alesina wrote a number of very influential articles on what have come to be known as Rational Partisan Theory (RPT).

In RPT models we essentially assume that we have a New Keynesian Phillips Curve and agents form rational expectations about what macroeconomic policy (the level of inflation) we will have after the election. What causes the political business cycle is essentially “election surprises”.

Let me illustrate it. We assume we have two political parties. The first party (“right”) favours a macroeconomic policy that will ensure 2% inflation, while the other party (“left”) favours 4% inflation. Lets then assume that the “left” paty is in power and delivers 4% inflation in period t, but that we will have general elections in period t+1 and that there is a 50/50 chance which party win the election. That means that the rational and risk-neutral economic agent would expect on average 3% inflation in t+1.

That would mean that even though the left government delivers 4% inflation labour unions will negotiate wage contracts on an assumption of 3% inflation (rather than 4%). This will cause a drop in unemployment after the elections if the left party wins as we will get an upside surprise on inflation, which causes real wages to drop. On the other hand if the right party wins it will deliver lower (2%) inflation than expected (3%), which will cause an increase in real wages and cause employment to drop.

I must say that I always found the Rational Partisan Theory extremely interesting and I believe that Alesina’s models (and other similar models) move thinking about Political Business Cycles forward compared to the economically naive models of William Nordhaus. However, other being based o somewhat of a caricature of the “left” and the “right” I early on realized that there was one major problem with Alesina’s models and that was the way they complete lack any proper discussion of monetary policy rules.

Thinking about Rational Partisan Theory without ignoring the Sumner Critique  

A the core of the problem with the early RPT models was that they essentially ignored the so-called Sumner Critique. What Scott Sumner is saying is that if we have a central bank that for example targets 2% inflation then the budget multiplier is zero. Hence, if a “left” government eases fiscal policy to push up inflation to 4% then the central bank – given its mandate to deliver 2% – would simply tighten monetary policy to offset the impact of the fiscal easing on aggregate demand so to ensure 2% inflation.

Hence, if the central bank is fully credible the rational economic agent would always expect 2% inflation in t+1 no matter who would win the elections. This of course means that there might be elections surprises, but there wouldn’t be any inflation surprises.

Furthermore, if the government is not able to set inflation (as the central bank has the final word on aggregate demand and inflation) then there would essentially not be any reason why left and right should difference on this issue. Why would a left party ease fiscal policy when it would know that it would just be overruled by the central bank?

So in my view what we need is essentially a Rational Partisan Theory that takes the monetary policy rule into account and takes into account whether this policy rule is credible or not because if the the policy rule is not credible at all then we are back to the Alesina model. On the other hand under a credible policy rule the dynamics in the model is completely different than in the early Alesina models.

Similarly it is not unimportant what kind of policy rule we have. Take the example of Denmark and Sweden. In Denmark we have a fixed exchange rate policy, which means that the Sumner Critique does not necessarily apply – fiscal easing might increase aggregate demand and inflation – while in Sweden where the Riksbank has an mandate to ensure 2% it is more likely that we will have “monetary offset” of fiscal easing.

This means that if we want to test Rational Partisan Theory we could do it by comparing the development in countries with different monetary policy rules. Similarly – and this I think is highly important – we need to look at financial market developments rather than macroeconomic developments.

Exchange rates and Rational Partisan Theory  

This brings me to what really has caused me to write this blog post. This morning I had a talk with a colleague about how parliament elections could impact exchange rates and as we where talking I realized that the view presented by my colleague essentially was a Rational Partisan Theory model in an economy with a floating exchange rate and an independent and credible inflation targeting central bank.

I want to sketch that model here and what we are interested in is figuring out is how elections influence the exchange rate development ahead of and after elections.

As we assume that the central bank has a credible inflation target – accepted by the two parties (left and right) – it makes little sense to think of different political preferences for inflation. Instead I think we should think of the economic-political differences between “left” and “right” as the “ideological” view of fiscal consolidation.

So we start out in a situation where there is a budget deficit. Both parties acknowledge the problem and see a need for fiscal consolidation. However, the two parties disagree on the speed of consolidation. The “right” party favours “shock therapy” to reduce the public deficit, while the “left” party favours slower consolidation of public finances.

I would here have to make an assumption because one could rightly question why the left would favour slow consolidation even though it should know that fast consolidation would not impact aggregate demand (and employment) negatively as e would have full monetary offset if the central bank is serious about achieving its inflation target.

My way out of this problem would be to assume that differences in policy does not reflect difference in preferences regarding the macroeconomic outcome, but there a need to signal a certain general attitude. Hence, we could argue that by arguing for slower fiscal consolidation the “left” party signals are more “socially balanced” fiscal policy, while by advocating “shock therapy” the “right” party would signal more “economic responsibility”.

So now we have our “model”: Floating exchange rates, a fully credible inflation targeting central bank and two political parties who differs over the desired speed of fiscal consolidation.

Lets now try to “simulate” the “model”.

A Scenario: Right party in power, right party is re-elected 

In this scenario we have a “tight” fiscal stance in period t. To offset the impact of inflation and aggregate demand we have a similar easy monetary stance. However, the valuation of the currency – whether it is “strong” or “weak” would depend on the expectation for the future monetary and fiscal stance.

If we a 50/50 chance of a left or right party win then the rational risk-neutral economic agent would expect a “neutral” fiscal stance (somewhere between “tight” and “ease” and that would mean we would expect a monetary policy that would also be “neutral”.

However, the day after the elections we would know who had won and if the right party win we now (assume) that we will get a more aggressive fiscal consolidation than if the “left” had won. As a consequence on the day the election result shows a “right” party win the currency should drop. The scale of the depreciation will dependent on the electoral surprise. If it is a major surprise then the currency move will be bigger.

Similarly if we have a “left” party wins we should expect to see the currency strengthen.

These results might seem counterintuitive to some thing that “isn’t fiscal consolidation great so shouldn’t it led to a strengthening of the currency?”. Well maybe, if you think of on the impact on the real exchange rate and we can easily think of a situation where swift fiscal consolidation leads to a real appreciation of the currency, but given the central bank is independent and committed to its inflation target the central bank will not allow any real appreciation pressures to led to nominal appreciation as this would undermine the inflation target.

We can therefore also use this this knowledge to think of the impact on other asset markets – for example the property market or the stock market. Without going into detail this kind of model would tell us that a “right” party win would cause stock prices to rally on the back of increased expectations for monetary easing.

Political Business Cycle theorists should focus on money and markets

This leads me to my conclusion: I believe that a lot of insight about Political Business Cycles (and business cycles in general) can be learned by starting out with an Alesina style model, but we need to incorporate monetary policy rules into the models.

Furthermore, while we probably can learn something of empirical relevance by looking at macroeconomic data I believe it would be much more fruitful to study the impact on asset markets – including currency and equity markets – to understand the Political Business Cycle. The advantage of using financial markets data rather than traditional macroeconomic data is obviously the forward-looking nature of financial markets.

Furthermore, we have well-developed prediction markets (such as Hypermind) for political events such as elections, which provide minute-by-minute or day-by-day odds on different political outcomes. Hence, we could imagine that the prediction market is telling on a daily basis whether the “left” or the “right” candidate will win. We can then test the impact of changes in these odds on the exchange rate (controlling for other factors).

This would be a simple way of test the kind of RPT-based exchange rate model I have sketched above and it would at the same time be a test of Rational Partisan Theory itself.

I am not saying that such literature does not exist, however, I am aware of very few studies that ventures down this road. So I hope this blog post can inspire somebody to do proper theoretical and empirical research based on such thinking.

Policy coordination, game theory and the Sumner Critique

Here is Alan Blinder in a paper – “Issues in the Coordination of Monetary and Fiscal Policy” – from 1982:

“Consider the problem of designing a car in which student drivers will be taught to drive. The car will have two steering wheels and two sets of brakes. One way to achieve “coordination” is to design the car so that one set of controls – the teacher’s – can always override the other. And it may seem obvious that this is the correct thing to do in this case.”

The student driver obviously is fiscal policy, while the teacher is monetary policy. If the student (fiscal policy) try to take the car (the economy) in one direction the teacher (monetary policy) can always step in and overrule him. This is of course the Sumner Critique – monetary policy will always have the final say on the level of aggregate demand/nominal GDP and hence the fiscal multiplier is zero if the central bank for example targets the nominal GDP level or inflation and that is even the case if the world is assumed to be Keynesian in nature.

However, even though monetary policy has the final say that does not mean that monetary policy will conducted in the right fashion or as Blinder express it:

“But now suppose that we do not know in advance who will sit in which seat. Or what if the teacher, while a superior driver, has terrible eyesight? Under these conditions it is no longer obvious that we want one set of controls to be able to ovemde the other. Reasoning that a stalemate may be better than a violent collision, we may decide that it is best to design the car with two sets of competing controls which can partially offset one another.”

Blinder here raises an interesting question – what if the central bank does not conduct monetary policy in a proper fashion wouldn’t it then be better to give the fiscal authority the possibility to try it’s luck. Blinder is of course right there is no guarantee that the central bank will do a good job – if that was the case then we would not be in this crisis. However, does that mean that fiscal policy can “take over”? Obviously not – even a bad central bank can overrule the fiscal authority when it comes to aggregate demand. The ECB is doing that on a daily basis.

Anyway, I really just wanted to remind my readers of Blinder’s paper. It is really not directly about the Sumner Critique, but rather Alan Blinder is discussing coordination between monetary policy and fiscal policy from a game theoretical perspective. Even though Blinder obviously as a lot more faith in “government design” than I have the paper is quite interesting in terms of the games central banks an governments play against (and sometimes with) each other. I find Blinder’s discussion highly relevant for particularly the game being played in the euro zone today between the ECB and European governments about monetary easing versus fiscal consolidation.

William Nordhaus in 1994 wrote a similar paper to Blinder’s about “Policy Games: Coordination and Independence in Monetary and Fiscal Policies”Nordhaus’ paper is equally relevant to today’s discussion.

It seems like the game theoretical literature about monetary-fiscal policy coordination has somewhat disappeared today, but to me these topics are more relevant that ever. If my readers are aware of any newer literature on this topic I would be very happy to hear about it.

PS the literature apparently not completely dead – here is a 2010 dissertation on the same topic by Helton Saulo B. Dos Santos. I have not read it, but it looks quite interesting.

Update: Nick Rowe has kindly reminded me that he and Simon Power actually have written a paper on the same topic back in 1998. Nick recently did a blog post about his paper. Nick interesting enough reaches the same conclusion as I do that in a Stackelberg setting where the government sets the budget deficit first and the central bank follows and determine NGDP we get a outcome similar to the Sumner Critique. Again this is not due to monetarist assumptions about the structure the economy (the LM curve does not have to be vertical), but rather due to the game theoretical setting.

Follow

Get every new post delivered to your Inbox.

Join 4,618 other followers

%d bloggers like this: