LaTeX RAQs

* This post will be occasionally updated.

Here are some Recently Asked Questions about TeX, entirely by myself, and some answers I found on the web.

Graphics & Tables/Floatings

Q: How do I place pictures inside a table?
A: Tips from LaTeX Matters
(This is useful for arranging pictures in neat format, as well as importing tables with nice formatting into LaTeX as pictures.)
In short, you can pretty much use \includegraphic within a \table environment.

Q: How to place wide tables in landscape?
A: Use lscape and pdflscape packages. See this.

Q: How to put all tables and figures at the end of the documents (as asked by many journals)?
A: Use endfloat package. See also these general tips.

Q: How to use table in landscape mode with endfloat?
A: 1.) Copy the file: efxmpl.cfg (which is located in the package
location: ...\latex\endfloat on my hard drive) to where the tex document is and leave it in that directory.
2.) rename the file: efxmpl.cfg to endfloat.cfg.
3.) Compile.
*** I haven't managed to do this successfully.

Q: How to vertically align numbers in different cells of the table?
A: Lots of suggestions on this, but I find the most flexible and simplest method is to use \phantom{}. E.g. to align 4.56 with (0.02)***, with 4.56 aligned with 0.02, we centre both cells, and use \phantom{***}(0.02)***.

EE432: Questions for Midterm Exams

Here are some questions I got from the mailbox. Check back for more.

In Fischer's staggered wage model, why do we set m = -v(-1)?

The objective here is to make output as stable as possible. With constant money supply, we found that output is given by 0.5(0.33(u-E(u|-1)) + 0.67(u-E(u|-2)). We know that the first part (u-E(u|-1))=v which the policy maker cannot do anything about. The current-period surprise v cannot be predicted by anyone. But the second part (u-E(u|-2))=v+v(-1) contains past shock v(-1) which has been observed by everyone. Therefore v(-1) is adding to the fluctuations of output unnecessarily.

How should we set m to eliminate this extra shock v(-1)? Because m and u enter the output in exactly the same way and we know (u-E(u|-2))=v+v(-1), it makes sense to set m such that (m-E(m|-2))=-v(-1) so that to cancel out v(-1). Setting m=-v(-1) accomplishes this goal precisely

Could you briefly summarize (in words) why money works in Staggered wage setting introduced by Fischer, Taylor and Calvo and why it does not work in Lucas'?

In Lucas (as well as Friedman), the starting point is the natural rate hypothesis which is the idea is that any change in monetary policy will translate into higher inflation, simply because nothing real or fundamental has changed in the economy. Any change in the quantity of money is a change in nominal quantity, that should not affect real quantities like employment or output. The exception is when people confuse these nominal changes with real changes, as in Lucas where agents sometimes interpret aggregate demand shock (which is nominal shock) wrongly as a relative price shock (which is real shock). When they make this kind of mistake, they may respond by producing more or less, which is why monetary policy is effective. But people only make mistakes when the change in monetary policy manages to surprise them (i.e. there is unanticipated demand change). Money does not work in this setting, in the sense that monetary policy cannot always surprise people, unless it is implemented randomly (and even so, we show in Lucas model that this will soon become ineffective too).

In (new) Keynesian models, we don't have an immediate pass-through from nominal changes to price adjustments, because these models assume price (or nominal) rigidity. Because prices do not fully adjust, real quantities such as output or employment must adjust instead to any aggregate demand changes. For example, after a negative demand shock (e.g. a monetary contraction), if the prices do not fall, lower demand will necessarily lead to lower equilibrium output. This holds true even with rational expectations (as we see in Fischer's model). Since a change in monetary policy is a demand or nominal change, this means that monetary policy is effective.

Questions from readings:
Lucas' Nobel Lecture, p.675, how does it follow from U'(n)=x that the equilibrium level of employment n will be a decreasing function of the rate of money growth.

The rate of money growth in this model is x. That n is decreasing in x follows from the fact that the marginal utility function U'(n) is decreasing in n (because of diminishing marginal utility assumption).

EE432: Key to problem sets 1 and 2

Happy revising for exams!

EE432: Reading pack for topic 3

The lecture note and the reading pack for topic 3 is now available for download.

Let me know if you have problems unzipping the file.

By the way, we will have to reschedule the class for February 25th to some other date. More update coming up.

EE432: Q&As for topic 1 and reading pack for topic 2

This is the first post for this semester, welcome to EE432!

Here are some questions I get for topic 1, which are based on Blanchard and Fischer textbook.

In the text book on page 161, What is the rate of return on money? Is it n or ((the money demand @t+1) - (the money demand @t))/(the money demand@t)?

The rate of return on money is simply the change in the 'value' of money over time. In this model, money is valued relative to only one good available in the economy. The price level P_t tells us how much that good is worth relative to money, or equivalently how much money is valued relative to good. Rate of return on money is then simply P_t/P_t+1, the deflation rate! If you hold money, and it buys more goods over time, money is effectively paying a rate of return.

Also, there is a statement "we can rule out non-steady-state paths in which the price level is falling at a rate greater than n".
Does this mean that if g is greater than n, the non-steady-state is impossible?
If this is the case, why the non-steady-state is impossible.
With g greater than n, the real balance @t+1 divided by the real balance @t is greater than 1. Since this is possible, that g is greater than n should be possible, isn't it?

The last sentence of your argument is wrong. Suppose that there is a non-steady-state equilibrium where g>n. You're right that the real balance per person will keep growing over time, because money pays a higher return than population growth. As real balance per person keeps growing, it will eventually exceeds 1. That implies that each young person will have to save more than 1 unit of good. That is impossible since he is only endowed with 1 unit of good. So our assumption that there is a non-steady-state equilibrium must be false. (This is called proof by contradiction)

On the same page, does the word " a monetary equilibrium" mean the equilibrium in the money market?

Not quite. In Blanchard and Fischer, a monetary equilibrium is an equilibrium in which money is used. This is in contrast to a barter equilibrium, in which everyone ignores money and use storage technology if available. (In this model, since trade cannot take place anyway, a barter equilibrium is really the same as an autarky equilibrium, that is you are on your own and never trade with anybody)

Thanks for the questions and keep them coming.

*********************

Here's the reading pack for topic 2.

Happy reading!

MABE 2010: First Post

Apologies for the delay.

**** Updated July 14th ****

Here are the lecture materials for download:


NB: Lectures 5 and 6 in previous years addressed Foreign exchange market, which this year will be covered by my colleague instead.

All the groups should now have their topics confirmed. Do let me know as soon as possible if you are not yet part of any group, or if your group is still in search of a topic.

Let me repeat what is expected of this assignment.

1. Key output includes (1) a 15-20 minute presentation by one or more representatives from the group, (2) two hard copies of the presentation materials for me and my colleague, and (3) a report containing the details of the research. After your presentation, please send soft copies of both the presentation and report to me as well.

2. The presentation should aim to be both informative and fun. The key aim is to engage your audience and educate them. I plan to hold presentations on the last 2 lectures of our class, but now realize that there's only one lecture on the last week, i.e. 28th. I will try to schedule 2 classes for that last week, so that all groups will have more time to prepare. The ordering of presentation will be randomized, and announced soon.

3. There is no strict requirement about the length of the report, as I place most emphasis on the quality and depth of research rather than length. For example, there's little point of including pages and pages of tables and diagrams if there's no discussion of what they mean. But if you really need a ballpark, previous years' reports range from 10ish to 30ish pages.

4. The research topics are pretty loosely defined, so you are free as a group to decide what angles to conduct your research.


Let me know if you have any questions.

EE432 2010: Revision

Here are some Q&As, from my mailbox.

Topic 4

1. What is the intuition of the positive relationship between income and demand for loan?

If you read the paper, Bernanke and Blinder, they say "The dependence on GNP ( y ) captures the trans- actions demand for credit, whch might arise, for example, from working capital or liquid- ity considerations." This means the higher income the economy has, the bigger are the firms, and the greater need for loans to finance working capital or the need for more cash to run the business.

2. Is the condition "R greater than 1 and diminishing marginal utility" enough for C1* greater than 1 and C2* less than R? How risk aversion involves in this?

"R>1 and diminishing marginal utility" are enough to infer that C2* is greater than C1* (see equation 3). But to ensure that C1* is greater than 1 and C2* is less than R, we showed in the lecture that we need gamma*u''/u' being less than 1 for all gamma. This is exactly the definition of relative risk aversion. So we do need risk aversion.


3. Is there a typo on Diamond&Dybvig p.408? And are you gonna include the stochastic t in the exam? Or we should focus on the model studied in class?

The mathematical question will only be based on what we covered in lectures. For essays or discussion, I set no ceiling, you're free to discuss whatever you learn from the readings. (which line do u see the typo?)


Topic 5

I get the concept that time inconsistency arises when preference before and after are different, but Im confused about the graph and relationship among pies.
1. For time inconsistency problem, can you elaborate more about the implication when L = 1/2(pie-pie star)2 , so it is the same as commitment case?
And after we arrive at pie = pie star, which leaves total loss fn to be only 1/2 (y-y*)2, what is the implication? Do you expect further explanation or just how to get there?
Im also confused about the relationship btw pies, expected pie and pie star, can you explain more on it. is it correct that the optimal solution is
for both society and cb to set pie and expected pie equal to pie star? is this the essence of it already? Also, can you elaborate more about how to interprete the graph?


Let me lay out the basics.

pi is the actual inflation. The solution for pi will have to be determined by an equilibrium condition. In this case, it is a Nash equilibrium, or in other words when we have both the central bank minimizing its loss function and the public holding rational expectations.

pi-star is a parameter, a number, that tells us what the ideal level of inflation is for the central bank. pi-star is in other words the targeted inflation. Now, in the lecture, I say pi-star happens to be what the society thinks of as ideal inflation as well, but this needs not be the case. If the society prefers some other inflation, then what this means is that the social loss function will be different from the central bank's loss function.

Expected pi, that's the expectation of inflation. Under rational expectations, it will have to be the same as actual inflation. This is the same definition of rational expectations that we work with since the first semester.

Your last statement, that the optimal solution entails that both the society and central bank should set pi and expected pi to pi-star is generally wrong. What do u mean by optimal? If optimal means that everyone is optimizing, then the time-inconsistency result precisely tells us that the optimal outcome could be a higher inflation than pi-star. That's our standard result (unless we consider some limiting case). But if you mean what is the outcome that yields the highest welfare, then yes, if the central bank can somehow make sure that pi=pi-star will be delivered in equilibrium, we have maximum welfare possible (assuming the society also thinks of pi-star as ideal). This will be 1/2(ybar-y*)2 by the way, in our original model.

Topic 6

(1) why the science of monetary paper really makes clear about cost-push and demand-pull? why the shock,u, can only can only cost push?

In the model we consider, 'u' is the shock applied directly to the Phillips curve, i.e. it's an inflation shock. That's why we call it the cost-push shock, or equivalently, it is a supply-side shock because it shifts the aggregate supply curve. The demand shock in this model is very easy to deal with, we just operate monetary policy to offset it. Remember I skip the details about the aggregate demand side because we assume the LM curve can be chosen to be anything we like….so even if IS shifts up or down, the LM can be chosen to move to offset it.

Intuitively, with demand shock, there is no interesting tradeoff. Negative shock hits, inflation and growth will be lower, so u want to expand monetary policy to boost demand. When there's a supply shock, that's interesting. Do you want a higher inflation, or a lower growth? Or a combination of both? How much?"

(2) in the case of commitment rule, is it always that k must equals to zero?

It doesn't have to be. 'k' is a preference parameter, so it depends on the central bank's preference. I let it equal zero for simplicity, and to highlight the key point here that even if k is zero, there is a gain to commitment (check that you understand why). In the standard simple time-inconsistency problem, if k=0 we don't have any time-inconsistency, and commitment or not it doesn't matter.

EE432 2010: Extra problems for topics 6 and 7

Download them here. There is no need to submit them, as these are just for practice.

For those of you resubmitting problem set 4, I left them at the BE office some time last week so you can pick them up.

As for extra tutorial class, we may not be able to use a room at Thammasat, as the exams already started. I will try to look for an alternative location. In the mean time, if you have questions, you can send them in right away, there's no need to wait.

EE432 2010: Topic 7 Materials

Presentation file and reading pack for our final topic is now available for download


Key readings are articles in the JEP symposium, contained in a separate 'JEP articles' folder.

EE432 2010: Topic 6 Materials

Lecture note and the reading pack for topic 6 is now available for download at:


As I mentioned, the key reading is the paper by Clarida et al, up until the part covered by the lecture note. Among other highly recommended readings (though not required) are
  • Blinder's "What can central bankers learn from academics and vice versa", from the perspective of someone who's been on both sides of the fence.
  • Bernanke's "Inflation targeting: a new framework?" published 3 years before Thailand imported the idea, which we still use to this day.
  • Woodford's survey article on the optimal monetary policy for the Handbook for Monetary Economics, which is the most up-to-date account of what the current research frontier is. This is quite advanced, so perhaps just skim through to get some general ideas.