Revised

Members present

Lord Hollick (Chairman)

Baroness Blackstone

Lord Forsyth of Drumlean

Lord Griffiths of Fforestfach

Lord Kerr of Kinlochard

Lord May of Oxford

Lord Monks

Lord Sharkey

Lord Turnbull

________________

Examination of Witness

Carlo Cottarelli, former Director, Fiscal Affairs Department, International Monetary Fund

Q108 The Chairman: Carlo Cottarelli, good morning and welcome to the House of Lords Economic Affairs Committee. As you know, I am sure, we are currently conducting an inquiry into the devolved finances of Scotland. In looking at that topic, one for which the arrangements are currently under negotiation between London and Edinburgh, we have taken some evidence about what happens abroad and in other countries that have already experienced this and have arrangements in place for devolved financing. You have written on the topic and have spoken widely on it, so we are very grateful that you have joined us today.

I wonder if I might start by asking you to perhaps paint the picture of the arrangements that have been put in place. Is there common ground between those arrangements? Which arrangements clearly have worked? What is the purpose behind the arrangements? Is there a set of principles that tend to guide these arrangements? If you could just give us an overview, then other Members will want to come in with more detailed questions.

Carlo Cottarelli: First of all, thank you very much for inviting me. It is an honour to be here. You ask about the general issue of fiscal decentralisation, and no doubt there has been a trend in the last few decades towards increased decentralisation. There are good reasons to decentralise or devolve fiscal autonomy. It is normally thought that the closer the delivery of services is to the population, the better they are, because proximity implies that there is more accountability, first of all. There is also a better alignment of incentives: if it wants to get better services, the local population has to pay more. So there are good reasons to decentralise.

At the same time, it is acknowledged everywhere, even in the most decentralised fiscal federations, that there is a need for a sizeable central Government to do things that it would be inappropriate to decentralise. There are things that are the essence of a state, a nation. Here, I am talking about defence and foreign affairs, for example. There are things you want to maintain centralised because of economies of scale. There are things you want to maintain centralised because you want to avoid competition, particularly tax competition and a drag to a lower level of the tax rate because of this competition. In general, there are spillover effects if you decentralise. There is an important risk‑sharing motivation as to why you want to maintain certain things at the central level and, of course, there are distributional reasons why you want to maintain things centrally, because, in this way, you will be able to redistribute income across regions, if that is a policy priority.

Finally, there is a very important and sometimes underestimated reasons why you want to maintain things at the central level, namely that centralising policies makes various parts of the nation more similar to one another. For example, when you centralise corporate income tax, you create a level playing field for investment.

There are good reasons why you want to centralise; there are good reasons too why you want to decentralise. If you look at the experience of fiscal federations in general —I am talking about fiscal federations because they are relatively more decentralised than unitary states—you find that, when it comes to revenue decentralisation, you have anything between 10% and 60% of revenues and anything between 30% and 60% of spending at the local level. In any case, the central Government never gets below 40% to 50% of spending and revenues. That is a broad picture. I will stop at this point.

The Chairman: Have you, in the work that you have done, seen any link between devolved fiscal powers and economic growth? We have heard evidence on both sides of that argument, and I wonder where the evidence that you have seen points.

Carlo Cottarelli: Let me first briefly summarise the theory. The theory will tell you that, precisely because there are good reasons to decentralise, you would think that some degree of decentralisation would increase economic growth because you would have a more efficient government. By the same token, there are reasons not to decentralise, so even in theory there would be a U‑shaped curve in terms of fiscal decentralisation being associated with faster growth. Nobody really knows, honestly, where the right point is. I have seen all kinds of result in the empirical literature. Some suggest that more decentralisation is preferable; others are of different views.

It is clear that certain things are necessary for decentralisation to work. If you decentralise, you need to make sure that the local or subnational governments behave properly from a fiscal point of view—they do not misbehave and they do not create a spillover effect with imprudent fiscal policies. That is essential. That is something I am pretty sure about. We have pretty sound results. On average, we have seen, in cases where there was decentralisation without fiscal discipline, things did not go well for the nation as a whole in terms of growth, because you end up in fiscal crisis, basically.

There is also an interesting paper that was recently presented at the IMF, based on the experience of Italy, which makes essentially one point. Fiscal decentralisation helps growth not only at the local level but at the national level if what the authors call the “social capital” is well distributed across the nation. Social capital, in the definition of the authors, is a behaviour that does not incentivise free‑riding. If you have some Governments that are free‑riders, then you had better not decentralise. It stands to reason that that is the case. This comes from Italy and it is related to the debate in Italy on fiscal decentralisation to the south. Some authors, in the north particularly, believe they are not fiscally responsible, especially as there is not much social capital in some regions of Italy. If there is no social capital, in that case, you do not want to decentralise because it would affect the growth rate not only of those regions but of the nation as a whole.

The Chairman: Having strong social capital is a very important prerequisite of making any decentralised structure work.

Carlo Cottarelli: Yes. You want to have social capital, and strong and fiscally responsible institutions. That, I am pretty sure, is a condition for effective fiscal decentralisation. Other than this, as you mentioned, you can find any kind of results in the literature when it comes to growth rates.

Q109 Lord Kerr of Kinlochard: Which are the best taxes to devolve to subnational governments and why?

Carlo Cottarelli: Let us start again from the theory and then go to the practice. The theory is pretty clear. You want to maintain at the central level taxes where the base is very mobile, because you do not want to get distortions in the allocation of resources and you want to avoid tax avoidance. You want to maintain at the central level taxes that are unequally distributed across the national territory, as long as you care about income distribution across regions. You want to maintain at the central level—it is another argument—progressive taxes, again, if you care about distributional goals. You want to maintain at the central level taxes that are very cyclical and are very much affected by the economic cycle, assuming that the centre is in a better position than the periphery to finance deficits in a downturn, which is a reasonable assumption, certainly in Italy and in Europe in general. That is one of the reasons why it has been argued that, in the euro area, you want to centralise some very cyclically sensitive taxes, because, in a downturn, the euro area treasury, let us say, which does not exist but potentially could exist, would be able to finance itself more easily than at least the weakest part of the euro area. This is the theory of what you want to maintain at the central level. Everything else you decentralise. Based on this, typically you want to keep at the central level corporate income and financial transaction taxes, because the base is very mobile and because at least corporate income tax is very cyclical. In practice, you will find that corporate income tax is centralised very often.

VAT is a bit different because, in principle, VAT is a consumption tax. One could argue that consumption is less volatile than income, for example. In practice, the most common form of indirect taxation is VAT. VAT is difficult to decentralise, because you break the debit‑credit chain, which is basically the definition of VAT. You would have to have borders if goods go to another region. That is why, in most cases, VAT remains centralised. There are some cases of decentralised VAT. The one that works relatively well is the one in Canada. India and Brazil also have decentralised VAT, but they are a bit of a mess. In India, if I remember correctly, services are taxed at the central level and goods at the local level. It does not work very well.

There is also the so‑called VIVAT, which is a proposal that has been made to make possible decentralization VAT. Basically, it says that all transactions among businesses are taxed—for example, at 20%. The final taxation is taxed at 22%, for example. That has not been tried. It is something that has been described in the literature, but you do not find it in practice. In principle it could work, but, again, it has not been tried. You find instead, at the local level, other retail taxes. In the United States, those are fairly common. While, in general, consumption taxation could be local, in practice, because it is VAT, it is kept central in most countries.

Personal income taxes are a mix. You typically have a strong central personal income tax, and in some cases you also have local taxation, as in the United States, in Canada and in Switzerland. In some cases, as in Italy, regions that are running deficits are allowed to have a small surcharge on personal income tax. When I was living in Rome last year, because healthcare in Italy is decentralised and the healthcare in the region of Rome, Lazio, was not very well managed—it was running a deficit—I had to pay a small surcharge on my personal income tax. In general, even in the most decentralised federations, you still have a large personal income tax.

What you definitely find completely decentralised in most cases—I would say in all cases I know—is real estate taxation. It is a perfect case. It is not very mobile—it is not mobile, if you want to say so—and it is not very cyclical, so that is kept at the decentralised level.

There is one last thing I want to say. Even when taxes are decentralised, it may be useful to maintain at the central level some components of the tax—for example, the definition of the tax base—because that helps tax administration. You can still maintain a single tax administration with economies of scale.

Lord Kerr of Kinlochard: In a country such as the United Kingdom, where large parts of the corporate sector operate in all regions of the country, would it make sense to devolve corporation tax? Can you think of examples where corporation tax devolution has worked well?

Carlo Cottarelli: No, not really, in the sense that I am not aware of any complete corporate income tax devolution. There are some countries that have subnational corporate income taxes: Switzerland, Canada, the United States. They have local corporate income taxes, but together with a central corporate income tax. When you have a decentralised corporate income tax, one issue is how you distribute taxes, because firms operate in various parts of the country. If a firm that is established in Delaware operates throughout the nation, how do you distribute the tax revenues? There are typically formulas that are used in these cases; the approach is called formula apportionment. In principle, you should have the same formula across the states of the United States, for example. This is not the case. There are differences, unfortunately, which create some complication, but that is basically the approach.

I do not know whether you have followed the debate at the international level on base erosion and profit shifting, and what the OECD has done. The critics who say that the OECD has not done enough argue that they should have pushed towards a worldwide formula apportionment approach. That was not the case. It is a complicated thing to do. That is another reason why corporate income tax is not very often decentralised. As I said, the US, Switzerland and Canada are, in general, a group of countries that we looked at in our book are the most decentralised federations.

Lord Kerr of Kinlochard: It is sometimes said in this country that Northern Ireland is a special case, because of very low corporation tax and strange regimes in the Republic of Ireland. Do you think there is a case for devolution of corporation tax to Northern Ireland?

Carlo Cottarelli: I do not know. I know the argument that has been put forward for Northern Ireland with respect to the Republic of Ireland. I do not know enough to see whether that is necessary or not. Without knowing much, one could say there is some degree of stronger competition coming from Ireland. On the other hand, one could say, in the modern world firms compete worldwide, so why pay particular attention to what the tax rate is in the Republic of Ireland?

Q110 Baroness Blackstone: What are the merits of sharing tax revenues, as against giving complete control to a subnational government? For example, it is proposed that the Scottish Parliament should retain all the revenues from tax on earnings but none of them from tax on dividends and savings.

Carlo Cottarelli: When one says “assign the revenues”, there are different approaches. One can say that the central tax is shared with various regions, meaning that a certain percentage of the total revenue goes to a certain region and another percentage goes to another region. In this case, this is just like a transfer. I guess what you are referring to is something that essentially implies that I do not assign 30% of the nationwide taxes to, for example, Scotland; I say whatever tax originates from Scotland stays there. In that case, that is something that has some features of a transfer, but also has some features of a full devolution of taxation. Clearly, you still have a single tax in the nation, which implies, for example in that case, that Scotland will not be able to change the tax rate. You do not have that advantage, if that is regarded to be an advantage, but you have the advantage of maintaining a central tax.

Like the devolution of taxes, if you do this kind of assignment of revenues, however, you have some cyclicality. If the region that receives the tax is willing to take that cyclicality, then fine. It is something in between. Being in between a transfer and a full devolution of taxation is something that has advantages of one thing and disadvantages of others.

We can look at, in practice, what happens for some taxes. For example, you mentioned the personal income tax. There are some at the local level. They are not very frequent and it is not very frequent to find very high proportions that are left at the local level. There are, however, exceptions. My own country, Italy, is an example. In Italy, there are two types of regions: there are ordinary regions and special statute regions. The special statute regions are allowed to keep up to 100% of the personal income tax. It is normally 90%, but the Valle d’Aosta, which is the region close to France, keeps 100% of the personal income tax. The other special statute regions—Trentino‑Alto Adige, the one close to Austria, Friuli-Venezia Giulia, the one close to the Former Republic of Yugoslavia, and the two islands—keep 90%. But that is not normally what you find in federations. This is a special arrangement that was made for very strong political reasons at the end of the Second World War.

Baroness Blackstone: What you are saying is that it is very rare for subnational governments to retain all taxes on income.

Carlo Cottarelli: Yes, all taxes on income is very rare.

Baroness Blackstone: There are no examples that you can think of.

Carlo Cottarelli: Well, there are some cases in which you have personal income taxes. The small country of St Kitts and Nevis has two personal income taxes. Other than this, I do not think it is very common to have 100%. It is very rare to have 100% of the personal income tax left at the local level. As I said, Valle d’Aosta is one example and some other regions of Italy get close to 100%.

Lord Griffiths of Fforestfach: How much fiscal autonomy would a state like Quebec have, in Canada, over something like income tax?

Carlo Cottarelli: I do not know exactly what the arrangements are for Quebec. Quebec, again, is a particular case because, within Canada, not all provinces are the same, but I do not know the arrangements in Quebec well enough.

Q111 Lord Monks: I will turn the conversation away from tax for a moment to the grants that come from the central authority to the regions. At the moment, in the UK the deal between London and Edinburgh is that there is only one variable in that grant. It is more or less fixed. The variable is population. We are aware that, in other countries, needs and the situation in the region get taken into account. It can get quite complicated, as we have learned from the situation in Australia. Do you have any good examples to give us about how the block grants are distributed across different countries, how they fund subnational governments and what criteria they take into account?

Carlo Cottarelli: First of all, again, there is range of approaches: there are transfers, but then there is revenue sharing; and there is devolution of taxation. In principle, there are better ways to transfer resources than just looking at the population. The most appropriate way in principle, which is used in Canada, for example, but also in Italy—in Italy in a very sophisticated way, although it is still at the beginning—is to look both at the spending needs and the ability to raise taxes. In principle, you want to do so, because you want to take into account not only the spending needs of a certain area but also whether that area is able to take care of those needs by itself.

I mentioned Italy because it is an interesting case. Canada has basically the same approach, but it is more simplified. In Italy, for local governments, as of this year, 20% of the transfers are based on a very complicated system that takes into account the population, the kind of territory that you have—whether you are a mountain local government or a seaside local government—the structure of the population and the demographics. It is based on a huge questionnaire that includes thousands of questions, to tailor the transfer to the spending needs and to the taxation ability of a certain area. The new system covers 20% of transfers this year. In 2016, the percentage will go to 30%, then 40% and then 55% in three years. The idea is that, in four or five years, all the transfers to local government will be based on this formula. In principle, it is a good approach. How sophisticated you want to be in tailoring the calculation of needs and the taxation ability depends on your preference and how much you dislike complicated things.

Lord Monks: I have just one follow‑up. Take the Italian example that you have spoken of. It sounds horrendously complicated and a recipe for disputes of a very serious kind between the different regions and Rome on these things. Is that how it works or is it relatively equable?

Carlo Cottarelli: First of all, I have one clarification. The system I described applies to municipalities, not to regions. Essentially, most of the transfers to regions in Italy are for healthcare spending, and that is based just on population and the demographic structure of the population. For municipalities, there is a complicated system. Whether it will give rise to legal problems, I do not know. It was introduced this year; it is too soon to say. But the formulas and the approach were agreed with the National Association of Municipalities, so they have ownership of the process. It is a cumbersome and complicated process but, after three years of discussions, it has been fully accepted by municipalities in Italy, so, in principle, they should not have problems with that.

Lord Forsyth of Drumlean: I wonder, without getting into the detail, whether you have a view on the stability of a system where the grant was distributed from the centre without any account of the basis of need, but solely on the basis of population and what had historically gone before. I am thinking specifically of the way the Barnett formula works in this country between Scotland, Wales and the centre, where the amount of grant—the baseline, relates to what the money was in the 1970s, and, since then, a proportion based on population has been added. If you move towards a system where you have more tax powers and you retain a grant‑based system that is solely related to population, do you have a view, based on your experience, of the likely stability of that?

Carlo Cottarelli: When you say “stability”, what do you mean?

Lord Forsyth of Drumlean: I mean the scope for argument between the various constituent parts that they have not been fairly treated, which then leads to further problems.

Carlo Cottarelli: First of all, I am not too familiar with the Barnett formula and its history.

Lord Forsyth of Drumlean: You are very lucky.

Carlo Cottarelli: Having said that, I would presume that these things have to be discussed and, in a way, agreed jointly. If there is an initial decision to shift to a new system, a new approach, a new formula, and if that is accepted, then parties should be willing to accept the consequences of whatever decision has been taken. I am probably not well informed. My understanding was that, in the case of Scotland, the initial decision was to have a neutral shift. That relates more to the devolution of taxation power, that it would be a sort of compensation for the grant. As to changes in the way the grant is calculated—this is just common sense—if you accept a move to a different system, you have to accept the consequences. It is difficult to find good examples, because this process of fiscal devolution has been quite slow over time, in Spain, for example, and in Italy, so you do not really have these huge swings or huge breaks in a single year from the use of a certain formula to another formula. My impression is that, everywhere, things have been changing but slowly, precisely to avoid this kind of controversy that would arise if you had a big change all of a sudden.

Q112 Lord May of Oxford: I would like to take us a little wider than the focus we have had up to this point on essentially Europe and things near us, partly because I have spent something more than a third of my adult life in the United States and I think it is a very interesting example of where the various states feel that their tax should stay there. They have, right across the country, local taxes, very local taxes, state taxes and then the national taxes. It is a system that has things wrong with it, but a lot of things right with it. Some of the things are the inequities that arise in a system like that. If you contrast the finance available to the southern states in general with what is available in the north‑east, which is where I was living, around Boston and so on, there is a big difference in the tax revenues for running things. There is also quite a difference, not so much in the money itself as in the attitudes. You go to the north‑east and you go to the western parts around Caltech and so on, and you have people very focused in interesting ways on things like science, which is something that is a vexed issue in this country at the moment in many ways, and yet other parts are quite different. I wonder what your view of that is, because that is much less homogeneous and has lots of interesting examples of the compromises that are made between keeping the tax where it arose and distributing it in an equitable way.

Carlo Cottarelli: This is a preference that has been expressed for historical reasons by the United States. What you are saying is exactly true. The United States is one of the most decentralised federations. I would mention one more thing. The United States, in our sample, is the only country that does not have equalisation transfers across regions—money that is given from the centre to the states to equalise income. This does not exist in the United States.

Lord May of Oxford: That is because the states are raising it themselves, largely.

Carlo Cottarelli: Yes, not only that, but also because nobody wants to have the kind of redistribution of the income that you would have through these horizontal transfers or through the transfers that come to the centre to equalise income. There is still quite a lot of cyclicality in transfers in the United States. If a state of the United States, say California, is hit by a recession, its net transfers to the Government decline, but that is simply the by‑product of still having a large portion of revenues centralised because they are cyclical revenues, so you still have transfers to states that go up and down with the cycle. If a state has economic troubles, it gets relatively more money from the centre, but it is because of the operation of the so‑called automatic stabiliser, because of the centralisation of highly cyclical revenues. Whether this is good or bad, again, it is a political choice whether states want not to have redistributional grants, but the United States is the exception in our sample.

Lord May of Oxford: It is a nasty system in a way, in that it perpetuates inequalities.

Carlo Cottarelli: Yes, by definition. There are no equalisation transfers.

Lord May of Oxford: All the best universities are in the wealthier states, and so on. It is not coincidental.

Carlo Cottarelli: As I said, even in the United States, there is still some redistribution. Because personal income tax is progressive, the richer states contribute more to the federal budget than the poorer states, but it is a by‑product of the centralisation of personal income tax. Again, it also has cyclical features that are nice from that point of view, but it is not the same kind of equalisation transfers that you get in other federations, not to mention, of course, the unitary nations.

Q113 Lord Griffiths of Fforestfach: I wonder if I could make a statement and then ask Carlo Cottarelli a question. I was fascinated by what you said earlier about the paper given at the IMF on social capital. I just wonder if our secretariat could have a reference to it to read it. It reminded me of something I have quoted in many talks over the years: a study by Edward Banfield of a small town in southern Italy done at Harvard in the 1950s called The Moral Basis of a Backward Society, showing that, with the culture and institutions of so many small towns in southern Italy, it would have been impossible to have tax devolution; they just did not exist. I would be very interested to see if there was any relevance of that—I suspect not, but there may be, in a way—to Scotland.

My question is that, for better or for worse, we are in the process of devolution in the United Kingdom, not just to Scotland but to Wales and to Northern Ireland. I just wondered, from your vast knowledge of the literature and practical experience, whether you think there are any lessons that you have observed and we could usefully learn from either the research or the experience.

Carlo Cottarelli: First of all, let me clarify. The paper I mentioned was presented in a seminar at the IMF. It was written by three Italian professors. One of them now happens to be teaching at Duke University in North Carolina, so that is why he came to present the paper at the IMF. I can provide the reference. It is a nice paper.

On the general lessons about devolution and the conditions to make it work, I would just repeat what I said earlier. A critical condition is that constraints—fiscal rules or any other form of constraint—are put in place to avoid the deficit bias that would otherwise arise at the local level. Because you share the common currency, in principle this may create expectations of bailout from the centre in case of problems, so it is essential to have good fiscal rules at the local level. You also need to maintain certain things at the central level. I mentioned earlier certain taxes that should remain at the central level. It is also essential to have institutions that allow dialogue between the central Government and subnational governments so that there is co‑operation and exchange of information.

Then, also—it may look trivial, but it is not at all trivial—there is the accounting system. In Germany, the Länder do not have the same accounting system. It is very messy. In Italy, until this year, the regions did not have the same accounting system, which meant that, if you wanted to compare how many subsidies to the industrial sector Lombardy was given with respect to Piedmont—I tried to do it myself—it was very difficult, because the balance sheets and accounting systems were different. I do not think this issue will arise in the case of Scotland and Wales, because presumably you will maintain the same single accounting system in principle, but it is something that needs to be maintained in the future. This is my first reaction to the preconditions for good devolution. I would certainly mention the use of fiscal rules or some other form of constraint, the accounting, the dialogue and maintaining at the central level stuff that should be maintained at the central level.

Q114 Lord Sharkey: Carlo Cottarelli, the Smith commission recommended a principle where there should be no detriment to either nation as a result of UK Government or Scottish policy decisions post‑devolution. We heard from one of our witnesses, a Mr David Phillips of the Institute for Fiscal Studies, who said, “There was an event held recently in Edinburgh by a guy from the IMF who looked at 13 federal tax systems and devolution systems. He commented, when asked about these issues, ‘Wow, I have never seen any other country try to have a “no detriment” principle’”. That may not of course have been you, but I wonder whether you would agree with that statement and whether in particular you think that the no detriment principle could work.

Carlo Cottarelli: My understanding is that there are two no detriment principles. One is, first, for the initial grant, and that is pretty straightforward. The other one is definitely more complicated, to the point of being in practice almost impossible. There are so many kinds of spillovers that, if you want to offset anything, in a way it is self‑defeating to the goal of fiscal decentralisation. I always thought that this second part of the no detriment principle was to be interpreted more as a general guideline, to make sure that the two parties talk and discuss things that affect each other. In a way, it is similar to what I was saying earlier, that one important condition for successful devolution is that there are fora for discussion of common issues. I would interpret the no detriment principle as something that does not need to be interpreted in a legal way and to the last penny, but as something that is a guideline and strongly encourages co‑ordination and discussion.

Lord Sharkey: Does it not seem to you, on the face of it, almost a logical impossibility?

Carlo Cottarelli: I would say more than logical—a practical impossibility, yes.

Lord Turnbull: In 1977, I was working in the IMF and I was sent on a mission to the Netherlands Antilles. The purpose of this mission was that the two islands, Aruba and Curaçao, would not talk to each other. They were in a fiscal federation. They would not talk to the Netherlands Government. The only person they would talk to was the IMF. All I can say is relations with Scotland are not great, but they are not that bad.

Lord Forsyth of Drumlean: Yet.

Lord Turnbull: We come back to this question of borrowing, which Lord Griffiths touched on. There seem to be two philosophies. One is that the federal Government sets a borrowing strategy—in our case, it is reaching a balance by a certain period—and the separate devolved authorities should broadly follow that and make a proper contribution to it; then you add in a bit of borrowing, to add a degree of flexibility to deal with cash-flow difficulties or whatever. There is a second philosophy that says these entities should stand on their own two feet; they can borrow on their own credit rating; the debt they issue is not guaranteed by the central Government; there is no bailout, and that should be sufficient. In the case of a state in the US, which model do they fit into? Does a state have free borrowing but run the peril that it could go bankrupt?

Carlo Cottarelli: In none of the federations that we looked at did we find the full implementation of your second approach. The second approach says, “Let the market take care of the problem. If a subnational government borrows too much, the market will penalise it; it will pay higher interest rates and this will put pressure on public finances. Everything will be fine.” None of the federations we looked at uses this approach, essentially because it is too risky. There is always the possibility that the market does not provide enough discipline and you end up, at the end, exposed, with the need to bail out a certain part of the nation.

The United States uses a mixed system. On the one hand, as you know, there are fiscal rules in almost all the 50 states. Most of them, I would say, are self‑imposed. The point is, as I have always argued, they are self‑imposed historically, but, had the states not had self‑imposed rules, they would have been imposed from the centre. I said “most”, but, no, all states have self‑imposed rules in the United States. In general, in federations, you find rules that are imposed from the centre. In the United States, you do not have rules imposed from the centre, but you have self‑imposed rules that are exactly equivalent. At the same time, you have the no bailout approach, which, in the United States, has worked pretty well. Since 1840, there has been no bailout of states from the federal Government. One could argue that the exception was the District of Columbia in the 1990s, but the District of Columbia is not a state. There have been bailouts of municipalities by states in the United States, but, in general, the approach has worked well.

Personally, I would not go for a system that just relies on market discipline. It is too risky, and it is not what we found in federations. At the other end of the spectrum, you have the approach, which is not very common but is found somewhere—I cannot remember where, but it is found—of having almost no discretion at all: every year, the budget of the subnational government is approved by the central Government. That is rare, except for cases in which the subnational government got into trouble and it essentially had to request support from the centre. From then on, for a few years, you have this vetting by the central Government.

Otherwise, the most common approach by far is fiscal rules, mostly with a ceiling on the deficit and a ceiling on the debt‑to‑GDP ratio. There are difficulties also with this approach, mostly related to cyclicality. It is a well-known problem, because it is difficult to have cyclically adjusted targets for subnational governments. It is very difficult to have cyclically-adjusted targets even for the national Government, and even more for subnational governments. There are various solutions: rainy‑day funds, for example. One could also have cyclical transfers from the centre.

Lord Turnbull: What if it happens, though, that a devolved authority has a rule, but it is a rule that the central authorities strongly disagree with?

Carlo Cottarelli: As part of the devolution agreement, there should be rules for the local government that are accepted by the central government. The euro area is a somewhat different case, but, from the beginning of the introduction of the euro, the members of the euro area agreed to accept rules that were agreed by everybody. So it is not enough to have any rule at the local level. The rule, at the central level, must be approved by the nation as a whole.

Lord Turnbull: That is interesting. The euro zone is not really a very good example. First, they cheated in order to get in, and then they ignored the rules, and even some of the bigger countries like France were the leading offenders in saying, “We are not going to follow these rules”.

Carlo Cottarelli: You can say that the European Union is not even a state; it is not even a nation. Even in nations, of course, there are difficulties in practice. Brazil relied on fiscal rules that were not working for a while. Then, in the last 15 years, the rules existing for the Brazilian states’ subnational governments were working very well. There is always tension, but it is necessary to have rules. Very often, if a set of rules does not work, then you end up bailing out the local government, but the counterpart of this is the acceptance by the local government of tighter rules.

Q115 Lord Forsyth of Drumlean: Following on from that, what kind of institutional arrangements do you think would be necessary where you have devolution of fiscal powers?

Carlo Cottarelli: It is linked to what I said earlier. Again, fiscal rules are part of institutional arrangements. You need to have common accounting and bodies or institutions where you can discuss fiscal policy issues or spillover issues. You may also want to discuss in those fora the nationwide fiscal stance, although, ultimately, the responsibility of the nationwide fiscal stance rests with the central Government.

Lord Forsyth of Drumlean: How does that work? I am thinking, obviously, of the situation with Scotland in particular, where, clearly, you need to have these institutions to work. In fact, in your Edinburgh presentation, you pointed this out as being a factor of a federal system. Of course, in this context, we do not have a federal system, but we have a Government that is dominant and actually wants to leave the state, and we have those tensions. Is it practical to establish institutions where one party wants to leave the state and that particular party may have a completely different view of macroeconomic policy?

Carlo Cottarelli: It is clear that the UK situation is a bit different from the others. Typically, one has the federal Government and then regional governments, whereas here, in my understanding, you would have a Scottish and a Welsh Government, and you would not have an English Government, I guess, so it is a bit awkward. It is an ad hoc situation. I do not think, however, that this would prevent the setting‑up of institutions or commissions, whatever you want to call them, to discuss common problems. In any case, as discussions of this sort, perhaps at the quarterly level, with exchange of information, become more frequent, I think it would be necessary.

Lord Forsyth of Drumlean: You do not think it is necessary to have a common belief in the state to achieve that—you can set up the mechanisms, but to make them work.

Carlo Cottarelli: Clearly, you must share the goal. If you set up a commission because you believe common discussion is useful, there must be a common belief in the usefulness of the institution. I wanted to add one thing. Among the institutions that may be helpful are fiscal councils. You have a good tradition—it is relatively short, but still important—with the UK fiscal council, the Office for Budget Responsibility. A similar office would be useful also for Scotland. Perhaps that reflects my bias towards fiscal councils. They are not a panacea, clearly but tThey can help. I have seen experiences of fiscal councils that are not very effective, but I do not think they can cause any harm, and they can be helpful.

The Chairman: Following on from that, how would a fiscal council work?

Carlo Cottarelli: Do you mean how it would work in terms of mandate, for example?

The Chairman: It is established and agreed upon by all parties, I presume.

Carlo Cottarelli: I would assume that, for a fiscal council for Scotland, first of all, there would be an agreement with the UK Government to the establishment of a fiscal council. Then there would have to be some, presumably Scottish, legislation to establish it. That legislation would also clarify that the fiscal council is independent, because that is basically the precondition for its work as a fiscal council. There would have to be a mandate for the fiscal council—to do what? Normally, a fiscal council looks at, first and foremost, the application of fiscal rules, whether the rules are applied properly. The more general mandate is to follow fiscal developments in a certain area. The Scottish fiscal council would look at fiscal developments in Scotland. Presumably, the Scottish fiscal council would also have a dialogue with the Office for Budget Responsibility. That is the way I think it would work.

Q116 The Chairman: Can we come now to the transparency of the settlement? There is a certain amount of high‑level understanding of the Barnett formula, but actually the working parts of it are fairly opaque. Reference has been made in previous questions to the potential here for discord and disagreement, particularly if the objectives and goals of the various parts of the United Kingdom are slightly different. If the information around the funding arrangements is not transparent, then is it the case that you are creating the potential for a great row, for discord? Is that what happens in other countries? How do other countries approach this problem?

Carlo Cottarelli: Transparency is critical. You mentioned some aspects that are not entirely clear. I think there should be transparency about the arrangements. When it comes particularly to funding arrangements, there should be full transparency, because otherwise you get into controversies. Again, looking at the experience of Italy, this has not to do with the funding; it has to do with who is responsible for what. In 2001, the Italian constitution was amended to give more responsibility to the regions. It was not very clear, unfortunately. There were some areas of shared responsibility and, because of this, it was not very clear whether the central Government was allowed to do certain things. The result of this is that now 80% of the cases pending at the constitutional court are controversies between regions and the central Government of Italy. Now, this part of the constitution is being changed once again, as part of the reform the Renzi Government is putting forward, to make the system more manageable. When it comes to the relationship between the centre and subnational governments, the most transparent you are, the better.

The Chairman: Can you give us an example of a set of arrangements, with transparency and the publications of guidelines and things like that, that in your view works very well? Which countries have managed to achieve the gold standard?

Carlo Cottarelli: I do not know whether it is the gold standard, but Canada works pretty well. Of the federations that we looked at, it works relatively well. We looked at many arrangements in developing countries, but, again, I do not think those are good examples. Italy is not a federation, anyway. It is not even included in our sample, but, again, it certainly is not a good example. I would say, in Canada, the system is working relatively well.

Lord Griffiths of Fforestfach: Canada is a country of enormous size. If you think of a country like Switzerland, would you say it works quite well in Switzerland too?

Carlo Cottarelli: Yes.

Lord Griffiths of Fforestfach: It is not a function of size, in other words.

Carlo Cottarelli: Switzerland also works relatively well, yes. I do not know what extent of fiscal devolution you want to have. Along with Canada and Switzerland, the United States also works relatively well. This system of having essentially a clear no bailout approach, coupled with fiscal rules that are in principle self‑imposed but would be there anyway, works relatively well, and certainly better than cases like Argentina and Brazil. In my view, this probably has to do not just with the institutions that link the centre with subnational governments, but with a host of other factors. We know that, in almost all federations, with few exceptions, most debt is at the central level. You end up in a situation in which, in any case, the central Government represents a large amount of fiscal policy. In the sample that we looked at, on average, 25% of public debt was local and 75% was central. In a case like Switzerland, public debt is a bit larger than central debt, so it is very decentralised, but it works well.

The Chairman: In those countries that you have mentioned, are the detailed guidelines of how the settlement is made published and is the settlement itself published reasonably early after it has been agreed?

Carlo Cottarelli: When you say “the settlement”, do you mean the agreement?

The Chairman: Yes.

Carlo Cottarelli: In cases like Switzerland, this comes from history, so I do not think there was even an intention to have a final settlement for ever. These are things that have been built incrementally.

The Chairman: Thank you very much indeed, Mr Cottarelli, for joining us today, for giving us such insightful answers and for sharing your extensive knowledge of how this works around the world. Our challenge is to make it work here, from scratch. Thank you very much indeed.