GTAP Virtual Seminar Series, Vol 3, No 1 (2022) – Finance in a global CGE model

So let me just very briefly welcome everyone to 
this GTAP virtual seminar series. We are honored   to have Professor Peter Dixon with us today to 
present his work on Finance in a Global CGE Model   based on the paper that was published in 
the Journal of Global Economic Analysis   very recently. Still warm of the press; I 
think it was just last month that it was out   in it. There's also an analysis on the effect of 
financial decoupling between the US and China.   My name is Angel and I'll be moderating, and I 
think the only last bit that I would like to share   before I ask Peter to take over is that we have 
about 30 minutes for the presentation followed by   questions and answers. Please do type your 
questions in the chat box as they occur   so that we address them towards the end. And 
without further ado, Peter, the floor is yours.

Well thank you. Yeah, so you've 
got the title slide there.   This is joint work. You can see it's based on 
that paper that's given up, the citation is given   on the bottom of that slide. So it's joint with 
James Giesecke, Jason Nassios, and Maureen Rimmer.   I'll just say one thing about that paper, which I 
don't often say this about referees and editors,   but this was a paper that really benefited from 
the refereeing and editing process at the Journal   of Global Economic Analysis. So it was a credit, I 
think, to Roberto Roson and the referee. So it was   a paper that definitely improved in that process, 
so thanks to them.

Okay, so we go to slide two. All right, so this is what we did. 
We added a financial module to an 18   region, 57 commodity version of a GTAP 
model that had already been extended   by having year-on-year dynamics and 
sticky wage adjustments. That means that   in the short run, if you do 
something good for an economy,   there's an increase in employment rather than 
a change in wages, and then in the longer term,   as the labor market tightens up, wages 
go up, so the long-term benefit of a good   policy is in higher wages, and the short-term 
benefit is in terms of higher employment,   and that this version of this GTAP model source 
we've got industry specific capital, so we added   the financial module to that version of 
GTAP and we applied it to investigate   the effects of financial decoupling between 
the US and China.

So go to slide three. There we go, slide three, yes. Here, we are 
good thank you. Okay, so the starting point   for this work was Ianchovichina 
and McDougall's global trust   and we made three extensions of 
the of the global trust idea.   We introduced bilateral relationships. Those of 
you who are familiar with the global trust know   that each country lends money to the global 
trust and borrows money from the global trust,   but there's no direct bilateral relationships, 
but if we're going to investigate   a policy like financial decoupling between 
a couple of nations that are having sort of   an economic war, we want to have a model that 
can help us understand the effects of one country   discriminate, not only in trade, but in financial 
flows, financial relationships against another   country, then we're going to have bilateral 
relationships. We allowed for funds allocated   by region s to region r, to then flow to region k. 
What that means is if China lends money to the US,   then the US can take that money and use it to 
finance capital in the US, or it can lend the   money on to third countries, and that's kind of 
important to make best use of the available data   which refer to holdings by s of financial assets 
in r, not holding some physical capital in half.   And finally, we specified optimizing rate 
of return, sensitive behavior by financial   agents in their decisions on where to place their 
money, so that replaces the sensible, but sort   of non-economic idea of how, that's in the global 
trust, of how money is allocated between different   destinations.

All right, so if we built the 
financial module around an 18 region asset   liability capital (ALC) table based on data from 
the IMF, US and Chinese statistical agencies,   and GTAP. So the first question is what is an ALC 
table, and that's in the next slide, slide four. All right here's a three region version. 
So the diagonal components are values of   physical capital in each region. So that says 
that the US capital within the border of the US   at the start of 2015, or if you like the end of 
2014, it was 53.85 trillion, and then if you,   the off diagonals are values of foreign financial 
assets, and the columns and liabilities in the   rows. So the United States had financial assets in 
China worth 1.14 trillion looking down that row,   all right, or 19.97 trillion, sorry looking down 
the column, or 19.97 trillion in the rest of the   world. Now, I'm going to struggle a bit on 
this key idea. This is the most important   controversial theoretical idea, okay, so we're 
going to specify a financial agent for each region   who allocates the region's financial budget 
across alternative assets, that is, determines   the composition of the region's column in the 
ALC matrix.

Now, this financial agent in the US   has a budget, and the budget in 2015, 
the beginning of 2015, was 74.96   trillion dollars. That's the column sum there. 
And where does that budget come from? Well,   the agent collects the savings from 
people in the United States. It collects   money from other countries, from the 
financial agency in other countries,   so the financial agents in China and the 
financial agents in the rest of the world   invest funds with the financial agent in 
the US, and the financial agent in the US   can finance it, can use the money to 
finance the domestic capital, that is   let it out to industries, and can send it back 
to other countries and send it back to China,   or send it back to the rest of the world,   right? So all the investments in this model 
are done through these financial agents.   Now, what about this financial budget? All 
right, so that's the next slide, slide five.

All right, so here's the determination of the 
financial budget. Right, so the financial budget   for the agent in country are at the end of the 
year. The ones mean end of year and the zeros   means start of year, so the financial budget 
at the end of the year for the financial agent   in region r is the value of the assets at the 
beginning of the year. That's the capital stock,   that's the VK0 revalued by inflation, that's the 
VR plus the foreign assets that region r holds,   and that's supplemented during the year by the 
savings that take place in region r and the amount   of money that's borrowed or given to region r by 
the financial agents in other countries, which is   the FL1, that's the foreign liabilities at the end 
of the year. Take away the foreign liabilities at   the start of the year and you can see there, see 
the idea is to connect to all this financial stuff   with standard GTAP, and you can see there the 
standard GTAP variables are beginning to appear   like SAVE and VK and V(r) are all variables in 
standard GTAP.

Now given the financial budget,   what does the financial agent in region r actually 
do, and this is the next slide, slide six. All right, well the financial agent, no slide six, 
back one, back one, that's it, thank you. Well   the financial agent chooses the composition of 
the assets i's going to hold on behalf of the   households, and all the, in the domestic economy, 
and then the foreign economies, and so on, so it   chooses the composition of those assets that's 
going to hold at the end of the year, that's   that Z1(s,r), so it's choosing the composition of 
the column of the ALC matrix to maximize the CES   function.

Subject to that budget constraint is 
the budget constraint, it's six, the CES function   has got rates of return, you see the R(s,r) is the 
rate of return that financial agent r expects on   its investments in region s. Okay now, in 
percentage change form that produces the nice   little equation 7a on the bottom of that page, 
which says that the percentage change in region   r's holding of assets in s depends on the 
percentage change in the budget which depends   on how much savings been done domestically and 
how much foreigners want to lend to country r,   and it depends on a substitution term which 
compares the rate of returns that are expected   on investments in s, with average rates of 
return averaged over all the destinations   that r can have for its financial 

So that brings us the question,   where do these expected rates of return come 
from. and that's in the next slide, slide seven. All right, so the rate of return that agent 
r, let's say that the US financial agent   expects on investments in the US is another 
GTAP variable, that's RORE, which is the   expected rate of return on fiscal capital 
in region r and the rate of return that the   financial agent in the US to pay expects on 
its investments in region s. That depends   on the rate of return, the average rate of 
return that's expected on the portfolio of   investments by the agent in region in s.

that when region r sends funds to region s,   it does so through the financial agent in region 
s, thus the rate of return that r expects on these   funds reflects the expected rate of return 
on the portfolio managed by the agent in s.   Now, that equation 9 has got one more variable 
in it, which is that T(s,r) on the end, which   we can use that variable in simulating financial 
decoupling. So if we want to simulate policy   that reduces the ability of the US to invest 
in China, or the ability of China to invest in   the US, we can do that by negative shocks to the 
variable called T(s,r), so a reduction in T(s,r)   reduces the rate of return that agent r expects 
on its investments in region s. Okay, so the last   bit of the theory is on in slide eight, which is 
about foreign income flows, so go to slide eight. We've got to collect up receipts that country 
r gets on its foreign assets and that depends   on the composition of those foreign assets, 
which is that Z0(s,r) and it depends on how much   assets in country s earned per unit of asset, and 
that's CR(s).

Now then, the next equation 13 gives   the payments on foreign liabilities by country r 
and those CR(s) which the income flows per unit of   in investment in well CR(s) is the income flow 
per unit of investment in country s. Those   income flows are dominated by the variable NR(s), 
which is the net rental on capital in region s,   and that's again, another connection with standard 
GTAP, that's a variable that comes out of standard   GTAP, and then last of all, we include the 
difference between the receipts and the payments   that are made by r in r's net national product, 
which is income in GTAP notation, so all these   connections with the standard GTAP model. Right, 
now we go to the application which is slide nine.

Right, so we're going to do three 
simulations. So this is all to do   with US-China financial decoupling, which was a 
possibility that was being discussed in 2019 and   2020, and actually, we were commissioned to do 
this work, so it was a relevant policy idea,   so the first simulation is the US cuts assets 
entrusted to China by 50%, and we do that by   exogenizing the path of Z1, that's the end-of-year 
holdings by the US in China, Z1(China, US),   and endogenizing the path of that T variable. So 
we tell the model what's going to happen to US   holdings of assets in China and let the model 
tell us what has to happen to that T variable,   what has to happen to the expected rates of return 
for US investors in China. And then simulation 2   is the opposite way around, China cuts assets 
entrusted to the US by 50%, and simulation   3 is both things happen simultaneously, they both 
cut their financial assets in each other by 50%.   Now the simulation, all these 50% cuts 
are conducted relative to a baseline,   so there's a fairly bland baseline that goes 
out to 2025, and the cuts are phased in over   three years which are 2016, 2017, and 2018. 
Okay, so we look at some results in slide 10.

Okay, now here, this is simulation 1. This 
table, it might take a moment to get used to,   okay, you can see the exogenous shock, 
which is the US as an asset agent,   asset agent is in the columns, so the US as an 
asset agent in 2016, reduces its financial assets   in China by 20.63%, then by 2017, the reduction is 
37%, by 2018 it's 50%, and then it's held at 50%,   right. And so there's a reduction in US assets in 
China phased in over three years. Now what does   that do? Well it causes the US to redirect funds 
towards domestic capital and you can see that by   2025, US domestic, that's if you look at construct 
2025, that's the panel on the far right there,   so the US directs funds towards domestic capital, 
you can see by 2025 that US capital is 0.3%   higher than that; otherwise, would have been, and 
looking down that column the US column.

In 2025,   you can see that rest of world, US 
investments in the rest of the world,   are 0.86% higher then they otherwise would 
have been, and US wealth is increased by 0.14%,   and that reflects favorable macro effects, 
which we'll discuss in a moment. Now for China,   look at 2025. Again, China's financial budget has 
been reduced, Canada hasn't got the money from   the United States, so China's financial budget 
is reduced leading to reduced domestic capital   and foreign assets. So look down that 
China column in 2025, it's all negative,   and Chinese wealth decreases by 0.17% in 2025, 
reflect the unfavorable macro effects.

Now, what   are these favorable and unfavorable macro effects? 
Well that's in slide 11. Next slide, slide 11. Here we go, so well the US winds up with more 
physical capital because of the money being   brought back to the US, so that's in the third in 
the panel C, there. So you can see the US capital   is up by 0.20% and the Chinese capital is down 
by 0.39% that allows the US GDP to be up you can   see the 0.06 in the top panel there and China's 
GDP is down. The process of building up capital   is good for employment in the short run, 
and you can see US employment rises,   that's the blue line in the middle 
panel with the sticky wage assumption,   and then labor market adjusts, and wages are 
permanently higher in the US and employment   returns to control, and the opposite for China. 
All right, so we look at the next slide, slide 12. Yep, that's the one. Okay, now this is the, yeah, 
do we take a question? Someone wants to ask me a   question. Okay, I should take questions at the 
end, I think, okay.

So in simulation 2, China   cuts assets entrusted to the US by 50%, so it's 
kind of opposite, so these results show that the   reduction in Chinese assets in the US, it's phased 
in over three years, China redirects funds towards   domestic capital, look at the panel on the right, 
you can see that capital in China rises by 1.81%.   In 2025, Chinese investments in the 
rest of the world are up by 7.68%,   Chinese wealth increases 0.62% 
reflecting favorable macro effects.   The US financial budget is reduced leading to 
reduced domestic capital and foreign assets   and US wealth decreases, so you 
can see that's down by 0.72% in   2025 reflecting unfavorable macro 
effects, and then go to slide 13. And you can see that slide 13, those 
macro effects, look at slide 13,   that we can see that those macro effects are 
qualitatively just the opposite of the macro   effects in simulation 1.

China's capital is up, 
China's GDP is up, China experiences a short run   favorable employment effect, 
and long run favorable   wage effects, and for the United States, 
it's exactly the opposite. Now slide 14. Okay, now if you're able to follow all those 
numbers, I'm sorry you would be able to,   these present days are always difficult for 
lots of numbers, but you might have noticed   that simulation 2 looks like a lot bigger than 
simulation 1, so we'll ask this question – why   is the capital effect in China of withdrawal from 
the US that's simulation 2, six times greater   than the capital effect in the US of 
withdrawal from China in simulation 1? So,   let's look at those numbers.

You can see that in 
simulation 1, which was good for the US, the US   capital stock was up by 0.3. In simulation 2, 
which was good for China, Chinese capital stock is   up by 6 times more, that's 1.81, and there are two 
reasons. One reason is that the impact effect is   four times bigger for China in simulation 2 than 
for the US. What do I mean by that? Well, China   has a lot more money in the US than the US 
has in China, right, so we withdrew 50%,   so China is bringing back a lot more money to 
China in simulation 2 than the US was bringing   back to the US in simulation 1. That's one thing, 
and then the other point is that on this impact   effect is that China's capital stock is smaller 
than the US capital stock, so in percentage terms,   simulation 2 is doing a lot more for China 
than simulation 1 is doing for the US,   but that only really explained a four times 
bigger effect, and the answer is six times bigger,   so we looked for the other part.

This turned 
out to be quite interesting. It's all to do   with the US financial markets that could 
do more open than China's, so when the US   brings back money, if you like, from China, US 
financial agents have much more opportunity to   send it on to third countries, so less of it stays 
in the US, and similarly, when the US brings back   money from China and invests more in the US 
thereby reducing rates of return on capital   in the US, this openness means that 
foreigners, third country foreigners,   withdraw their money from the US, so the US 
gains a good deal less in terms of domestic   capital stock from bringing back money than 
the other way around.

So now, implication.   Now we know that China wins in simulation 
1, 2 sorry, China wins in simulation 2,   and simulation 2 is a much bigger, quantitatively 
bigger, than simulation 1, so China wins in   simulation 2, the US wins in simulation 1, 
put those things together in simulation 3,   and you'd expect China to be the winner, 
all right, and that's confirmed in slide 15. So we're going to slide 15. There 
we are, so that's simulation 3. 50%   financial decoupling by both the US and 
China, and so China winds up with more capital   the US winds up with, less 
capital China winds up with,   more GDP, US with less GDP, China has a short 
run temporary employment gain, and the US has   a temporary employment loss.

Okay, so I think 
we're nearly there, so let's go to slide 16. Okay now it in terms of macro effects, the 50% 
financial decoupling is just as important to   the effects on GDP and capital and wages 
and employment zones are just as important   as a 50% trade decoupling, but both countries 
lose in tit-for-tat trade decoupling. That's   a bit different. You see in tit-for-tat financial 
decoupling, there were winners and losers, but in   tit-for-tat trade decoupling where both countries 
impose tariffs against each other, or inhibit   imports in some other way, both countries turn 
out to be losers. Now if we add trade decoupling   to financial decoupling, well then the trade 
decoupling, more or less, completely wipes out   the gains that China got from financial 
decoupling, so you can see in that diagram,   China's basically reduced to, well that's the 
GDP effect, China's more or less reduced to   zero GDP effect.

Financial decoupling hurt 
the United States and trade decoupling hurts   it more, and so you can see that the 
United States winds up fairly far down   the page. That's the GDP effect, but all 
the other effects are pretty much the same,   as you'd expect. Okay so that brings me to what's 
to be done next, and that's the next slide. So there's areas for future development. 
There's obviously a lot more data work,   there's always more data work to be done, 
and it's pretty hard to find clients who've   got enough patience and money for data 
work, but there's always more to be done.   This analysis depends pretty much on the ALC 
tables, the asset liability capital tables,   and we worked very hard on the on the US-China 
corner of the tables, and we've got control totals   on foreign assets and liabilities from the IMF, 
and we filled in most of the rest of the table by   [inaudible], which means that while we're 
reasonably comfortable with what we're saying   about China and the US, the model is not 
really suitable at the moment to extend to   third countries, other countries, right, so 
there's more work going to be needed there.   Now another area which you know, Jason Nassios is 
working on, which is very important, I think, is   disaggregation of the financial instruments into 
loans, bonds, equity cash, special drawing rights,   and that's the direction that CoPS, the CoPS 
team has been taking their single country models,   right, and that once in those single country 
models, but it'll be wonderful to do it in the   GTAP framework, but in the single country 
models, we can last in a general equilibrium   modeling, start saying something sensible and 
understanding things about monetary policy.   And then another area which we've been 
investigating in our single country models   is recognition of multiple financial agents for 
each country.

In this modeling, we just had the   one financial agent, but we need multiple 
financial agents in each country, which is   the financial agents that are making financial 
decisions, households banks, non-bank financial   institutions, retirement funds, industries, and 
government, and in terms of the application that   we've just gone through, once we have multiple 
financial agents, then we could take account of   the idea that the Chinese, when they invest in 
the United States, buy mainly bonds, financial   instruments like that, and the United States, 
when it invests in China, does a lot more FDI,   invests in actual industries, so so 
once you have multiple financial agents,   that's the sort of direction that the 
work we're doing could be extended. Okay,   I think that brings me to the end, so we've got 
a time for questions. Okay, thanks, so thank you. Thank you, Peter, yes there was a question 
in the chat by Madan Ghosh, and yes a round   of applause for Peter, very enlightening. 
I know you went really quickly over this,   but it's still very nice presentation, 
thank you, and just let me jump into the   questions because allowing more folks to 
ask as well, what is driving the results?   Is it the marginal productivity of capital, 
an imperfect capital market by Madan Ghosh.

Okay, all right, so what's driving the results 
is when the United States invests less in China,   then there's more money to be invested in other 
assets. So the United States financial agent is   pretty biased towards investing in the United 
States, so there's a home biased, so the United   States financial asset finances more capital 
projects in the United States, which reduces rates   of return that are required on investments in the 
United States, and then that causes the financial   agents in the United States to reconfigure and 
invest in other countries. So you can see that   if the United States can't invest in China, it's 
a reconfiguration of the United States' budget   towards its own assets and towards assets in 
the rest of the world, and that reconfigures   rates of return all over the world 
and cause all the financial agents   in the world to reconfigure their budget.

their investments, so what drives the results   are these rates of return which drive 
physical investment which drives employment,   and wage rates and all of the things that you'd 
expect to happen in the general equilibrium model.   So I hope that helps. We've 
done a lot of work in the paper   on 'back of the envelope' explanations 
of where all the results come from. Peter I had a question on whether the crafting of 
the baseline on those earlier years from 2014 to   say 2019 would be helpful to calibrate 
Sigma, the elasticity of substitution.

Yeah, okay, so what we actually know are all 
those shares, so we've got a pretty good handle   on what the financial budgets, how they've 
been allocated by the countries, all right, so   one of the good things about 
so much financial regulation   is that it generates quite good data, right, 
okay, but Sigma, that substitution elasticity,   the sensitivity of the allocation of the 
budgets between different destinations,   the sensitivity of that to change the expected 
rates of return, we know very little about that.   Now presumably, it could, people know a 
lot more about econometrics than I do,   maybe they could start trying to estimate 
it, but it looks like a pretty difficult job,   so what we did was we did sensitivity analysis, 
so in the paper, we vary that Sigma between,   I forget we took it from one to five or something, 
I mean we varied across quite a big range,   and sure the results depend on that Sigma, 
but qualitatively, all the things that I said   about the results stand up pretty well to 

In that Sigma, you don't want to make   the Sigma too low because then the allocations 
would be more rigid than we think is realistic,   and you don't want to make it too high otherwise 
they'd be too volatile, the allocations, so   the Sigma in the range of about three seem to, in 
my intuition, was only my intuition, seemed to be   about right, but fortunately the results were 
reasonably robust to variations in that sigma.

And then the other will either suggest,   well there's another question, I'll save mine 
for later. It's actually, let me open this,   all right, so Charles Xiao mentions thanks for 
the presentation, professor. I have a couple of   questions. Does the model automatically handle 
balance of payments? What is the implication on   exchange rates, for example will China withdrawing 
money from the US creating a decline in exchange   rates and boost US net exports. No expectation 
is introduced to the model, right? Best, Charles. Okay, so I should declare that Charles was 
my very distinguished student, so thanks,   Charles, for that who also wrote a very 
good thesis about financial modelling.   Okay, balance of payments, everything that's 
done in GTAP carries through, right, so   when the United States brings 
back money from China and   reduces required rates of return 
in the United States and stimulates   investment in the United States, the balance of 
payments moves towards deficit as you'd expect,   okay, so any all those mechanisms about the 
balance of payments and the relationship between   investment and saving the balance of payments, 
all of that goes through, and you can   you remember that the monetary, the 
income flows are all taken care of too.   One of the slides went through the income flows 
and they all feed into the balance of payments.   Exchange rates, that's a tricky one.

The exchange 
rates, there's no nominal exchange rates, okay, so   if we, so that the real exchange 
rates are determined in the model,   so when a country does more investment, it's 
real exchange rate, it's price level, appreciate   real exchange rate appreciate, so when 
the US brings back money from China,   it's real exchange rates price level relative to 
the price level the rest of the world goes up,   but there's no nominal exchange rates being 
recognized or movements in normal exchange rates,   and that would be a direction 
that we should go, if we can make,   if we can bring in more of these monetary 
instruments into this international model.   In our single country models, we do the exchange 
rate and the breakdown between movements in the   exchange rate and movements in the price level 
in determining the movement in the real exchange   rate, so there's something for you to do, 
Charles, give it a go, all right? Okay, next.

There's a question about whether the results   of the 2015 to 2019 match with real data, so 
I suppose this is related to the baseline. Yeah, the baseline is broadly right, okay. We 
cunningly stopped because of 2019 because things   get difficult by 2020. So the baseline introduces, 
the baseline is right at the macro level, okay,   it's not necessarily right at the 
industry levels, but it's going to be,   it's roughly right at the macro level, but 
it's a pretty bland baseline. Of course   in another project, we're taking that 
validation whether things are right or wrong,   we're taking that validation very seriously, we're 
actually trying to move say from 2004 to 2014   and work out what happened to technology and what 
happened to consumer preferences and what happened   to required rates of return on investment, so 
in an historical period, and then introduce   trends, and all those variables into the into the 
baseline and see if we can make the baselines gel   more closely, not just at a macro level, but 
at any industry level with the data, but no,   the baseline in this exercise was pretty bland 
and just broadly, right, at the macro level.

Thank you, Peter. I think this is 
a long one, so please bear with me.   Thank you for the very interesting 
presentation. For these large capital flows, how   would these relate to typical shocks 
seen in a central bank-type macro model?   In those models, capital flows 
respond to changes in asset prices.   These are affected by quantitative easing monetary 
policy and shocks to the risk premium, among   others. I am thinking of the most recent prominent 
example of tapering of quantitative easing in the   US and caused large capital outflows. This was 
a monetary policy action.

Does this model try to   replicate these types of policy-driven dynamics? 
Also, related to the last question, capital flows   should respond to asset price changes and these 
necessarily change exchange rates are exchange   rates important in the simulations in macro models 
those would drive results particularly for trade. Yeah, that's a hard one. All right, so I can 
do tiny bits of that. A lot of that is actually   done in our single country models, okay, and 
you can, there are references to papers by   Jason Nassios and James Giesecke and 
colleagues, right, so all sorts of   interesting things in those papers about how, if 
the government dictates that retirement funds,   that we, to have more money in retirement funds, 
how that changes capital flows and whatever, now   one of the many weaknesses of this thing that 
we've, that I've just presented, was that there   aren't any nominal exchange rate changes, so 
there was no valuation effects on foreign assets   and foreign liabilities, right, so that, so until 
the model gets some movements or some modeling of   nominal exchange rates, a lot of the rest of 
that stuff just can't be done in this model.   Now, on quantitative easing and so on, well 
I'm just talking off the top of my head now,   but I think we could do something about that. 
I think that affects those financial budgets,   so I think we could relate quantitative easing 
to the financial budgets and then simulate   changes in the financial budgets, but it's very 
much, that's off my head, so yeah, so I think I   better move on to another question and hopefully 
before I get into more trouble on that one, okay.

Yeah, well we don't have one. We got a thank 
you very much and I also want to thank you,   Peter. I think we're also reaching the 
end of the period and I think rather   than taking another lengthy question, I'll 
just say thanks again for taking the time,   stopping by thank you all for the excellent 
questions and maybe we can just wish you a   great day in Australia and hopefully 
for good more tennis results as well. Okay, well thank you very much 
and thank you for inviting me,   and if anyone wants to send me a 
question about something, or some   comments or suggestions for further 
research, or tell me that something's   already been done or I should be looking at 
something or other, please do, all right.

Okay, so thank you very much. Thank you and 
so there's a question about the presentation   being available. It will be available on the 
website along with this recording as well   the paper of course is available at 
the Journal of Global Economic Analysis   with supplementary materials so that 
you guys can replicate this as well,   the results and the figures. So thank you very 
much, Peter, all right, thank you, see you bye..

test attribution text

Add Comment