Presentations

Audience

For each presentation, each student who is not presenting should prepare at least one question for the presenters. Submit your question to the disucssion thread at least two days in advance of the presentation, so that the presenters have time to collate the questions and incorporate possible answers.

  • Be sure your question does not duplicate other questions already submitted.
  • Be sure your question demonstrates that you made a serious attempt to understand the paper. Provide enough paper-specific context for your question. Supporting quotes (with page numbers) will often be helpful.
  • General questions that could be produced by simply reading the paper abstract will not earn credit. Your question should clearly demonstrate that you have read the entire paper.
  • Do not simply solicit the presenter’s opion. Ask paper-specific questions that involve serious thinking about the results in the paper. Your question should be about the scientific process that the paper represents.
  • You may explicitly refer to additional literature by citing it and explicitly discussing it, but do not expect the presenter to come up with unspecified “other studies”.
  • Do not ask questions that would require the presenter to do outside research. The presenter should be able to turn to their paper when trying to answer your question. You are looking for clarifications, not soliticiting opinions.
  • Limit questions to 250 words.

Presenter

If you will want to use my computer for your presentation, please send me your slides ahead of time. Additionally, be sure to share your slides with the class (using the class email list). Presentations should comprise 15-20 slides and last about 30 minutes. You will not read your presentation from your slides. Instead, your slides should be carefully developed to visually support what you want to say. (Do a practice run before presenting.) You goal is to introduce, summarize and critique.

Introduce:

  • Who are the authors? What are their institutions? Is there anything you feel we should know about them? Show us pictures of them while you present their bio.
  • What are the research questions in this paper?

Summarize:

  • What are the core theoretical propositions (if any)? (You may copy figures from the paper, but don't try to discuss too many figures.)
  • Where do any data come from? (Be detailed: If we want to replicate the study, is the data adequately documented?)
  • What are the key empirical methods used?
  • What are the key results? (You may copy tables from the paper, but do not try to discuss too many tables. When tables are dense, copy only a part per slide so that your slides remain readable.)
  • If your chosen paper is long and wide ranging, you will not be able to discuss everything in it. Instead, pick the most important parts of the paper to discuss (and explain what makes them the most important).

Questions:

  • What questions did you receive from your classmates? What answers if any can you offer? (Do not answer each question separately; that will take too long. Instead, group related questions, and try address each group of questions.)
  • What remaining questions do you have about the paper?

Critique:

  • Is the theoretical framework (if any) plausible? What questions do you have about it?
  • Is the data quality good?
  • Are the data adequately documented? (I.e., could you replicate this paper?) Can you find links to the data to share with the class?
  • Are the empirical methods chosen appropriate?
  • Is the analysis persuasive? Why or why not?
  • Are the results interesting? Why or why not?

Analytical Exercises

Introductory Concepts

  1. In class we showed that covered interest parity should always hold when capital mobility is high. We usually work with the approximation i - i* = F / S - 1; the exact expression is F/ S = (1+ i)/(1+ i*). Suppose the three-month interest rate on XYZ deposits is 3.1%/year. The three-month interest rate on USD deposits is 1.5%/year. Suppose the present spot exchange rate is USD 0.9 per unit of XYZ. Compute the three-month XYZ-USD forward rate, and explain your computation in detail.

  2. Unfortunately forward rate data are bit hard to find free online. But at least for the moment, UBC at the website http://fx.sauder.ubc.ca/CAD/forward.html offers a summary of forward rates at various maturities. Use these data for the following exercises.

    1. Enter the spot and forward exchange rates into a single column of your spreadsheet. For example, on 2014-01-27 I found reported the following exchange rates (spot and forward): [0.9010, 0.9003, 0.8997, 0.8990, 0.8972, 0.8953, 0.8933, 0.8850, 0.8724] for [Spot, 1 month, 2 months, 3 months, 6 months, 9 months, 1 year, 2 years, 5 years]. On 2009-11-24 I found reported the following exchange rates (spot and forward): [0.9472, 0.9472, None, 0.9472, 0.9471, None, None, None, None, None]. So many missing values is unusual for this site but was typical during the global financial crisis. For example, on 2008-11-05 I found reported the following exchange rates (spot and forward): [0.8687, 0.8679, None, 0.8680, 0.8685, None, None, None, None, None]. But on 2006-02-18 I found [0.8679, 0.8686, 0.8693, 0.8701, 0.8722, 0.8760, 0.8872, 0.8955, 0.9016, 0.9060]. Naturally, your numbers will depend on the date you access the webpage.

    2. Use your spot and forward rate numbers to compute the rest of the table you see at this website. That is, add the appropriate formulae (from the website) to your spreadsheet in order to compute the forward premium points on CAD (also called the swap rate) and the implied foreign interest rate differential. Note that you are not supposed to find the formulae in your book or notes: you are to use what you have learned to figure these out by reading this website. (The premium is just a scaled arithmetic difference, and FIRD is defined on the UBC page.) Please explain the formulae that you use.

    3. Produce a graph similar to the UBC graph of forward premium points but with one difference: use your computation of the implied foreign interest rate differential instead of the forward premium points.

    4. Based on your graph and data, how would you describe the market's expectations about how the USD-CAD exchange rate will behave over time? Explain carefully.

      Note

      The website formula already produces a percent value, so do not use your spreadsheet to do this a second time (e.g., by setting the cell formatting).

      Note

      Computations in the UBC table sometimes appear slightly off at short horizons. I inquired about this and was told it is due to rounding error in the exchange rate reporting. And actually, you can see that the data are rounded by looking carefullly at the accompanying graph, which is apparently based on the raw data.

      Note

      Do not fill your graphs with fancy “fluff” (like 3-D images). You may want to read the work of Edward Tufte, if effective presentation of data is likely to be part of your work.

      Note

      If you are going to speak of an “exchange rate depreciation”, make sure it is clear which exchange rate (CAD-USD or USD-CAD) you mean. It is generally better to speak, e.g., of a USD appreciation/depreciation.

  3. Your US company is expecting a payment from an EU firm of 1M EUR in 3 months. How can you hedge your exposure to exchange rate risk in the forward market? Consider the following scenario. The EUR-USD spot rate is 1.48. Your bank offers you a EUR-USD 3-month forward rate of 1.50. Your EUR-USD (3 month) expected future spot rate is 1.48. If you contract today what do you pay your bank on the settlement date? Does it matter if your expectations prove correct or incorrect?

    Your US company is expecting a payment from a CN firm of 1M CNY in 3 months, but since CNY are nondeliverable you will receive instead the US value of 1M CNY on that date. The Chinese forward market for CNY is restrictive and unattractive, but there is an active offshore market in nondeliverable forward contracts. How can you hedge your exposure to exchange rate risk in the nondeliverable forward market? Consider the following scenario. The USD-CNY spot rate is 6.82. Your bank offers you a USD-CNY 3-month NDF rate of 6.75. Your USD-CNY (3 month) expected future spot rate is 6.82. If you contract today what if anything do you pay your bank on the settlement date? Does it matter if your expectations prove correct or incorrect? (Be very specific about the details, so that it is clear that you understand the difference between an NDF contract and a forward contract.) You may find the follwing helpful: http://www.bis.org/publ/cgfs22fedny5.pdf

PPP Analytics and Evidence

  1. What do we mean by “purchasing power parity”? Under what conditions should purchasing power parity hold? (Give an algebraic analysis, including a discussion of the monotonicity and homogeneity of price indexes.) Which commodities would you expect to satisfy these conditions?

    Empirically, is purchasing power parity a good characterization of the relationship between exchange rates and relative price levels in the short run? In the long run? (Refer to emph{specific studies or data}, including data examined in your homeworks, to support your answers.)

    Relate the statistical notion of a random walk to the concept of expected purchasing power parity.

Other Questions for Review

  1. Countries have surprisingly varied policies toward providing free access to macroeconomic data produced with tax payer dollars. For example, detailed Canadian time series data are sold by Statistics Canada, while South Africa makes a substantial amount of data freely available. Suppose I need a time series for the CAD-USD exchange rate, and I do not have funds to buy it from StatCan. How could I use the data at http://www.resbank.co.za/economics/histdownload/histdownload.htm to produce this? Be sure to describe the following:

    • What data will you want to use?
    • How will you get the raw data into your data analysis application (e.g., spreadsheet)? (Do it for one series, and describe what you did.)
    • What would you need to do with the raw data to produce usable time series? (You are not asked to produce an entire series; just discuss this.) What could go wrong as you try to use 2 time series to produce one new one? (Do not worry much about that fact that that you cannot have data for every day; that is true of the existing series. Worry about matching them up correctly.)
    • Be specific about any computation you will do. (Illustrate with an actual compuation, using data from a single date of your choosing.)

    What features of international financial markets enable you to do this?

Other

  1. PPP You need to collect some data. You may use any source: just be sure to document your data precisely. The library has the IMF's International Financial Statistics as hard copy and on disk: the reference desk can help you get data from the CD. If you are interested in a particular country, get data for that country. Otherwise pick a country that starts with the same letter as your last name. For your country collect spot rate and price data. Collect a spot exchange rate (against the dollar) for 1973 and 2000 and calculate the gross depreciation for the period: S(2000)/S(1973). Also collect prices P(1973,b) and P(2000,b), where b is the common base year. Use these to compute the gross inflation over the period: P(2000,b)/P(1973,b). Also collect prices P(1973,b) and P(2000,b), where b is the common base year. If your gross inflation calculation differs from your gross depreciation calculation by an order of magnitude or more, you have probably made a mistake in your data collection.

    Note: If you have only P(1973,b1) and P(2000,b2) where b1 and b2 are two different base years, you will have to get another piece of data, P(b2,b1), so you can calculate P(2000,b1)=P(2000,b2) x P(b2,b1). Don't forget your price indices will be expressed as percents: e.g., a reported value of 270 means P=2.7.

    Taking your country as the domestic country, compute PPP(2000)/PPP(1973), where PPP(t) is the relative price level at time t: PPP(t)=P(t)/P*(t) Here is the information for the US P(1973,1982)=0.45, P(2000,1982)=1.72. You are interested in PPP(t) changed over time compared to how the exchange rate changed. Turn in a brief write-up . Include a precise statement of where you got your data. (Precise enough to ensure replicability.)

  2. EViews HW: Replication of Mark Figure 3.3

    Replicate (approximately) Figure 3.3 in Mark. (Assume the innovation to the bubble is standard normal.) You will need to use what you learned in the last homework. You may need to ask some questions on the list as well ... Conduct unit root tests on your fundamentals solution and your bubbles solution. What do you learn?

    Estimated time for completion: 2 hours.

  3. Do the Chapter 3 end-of-chapter problems in Mark. Use the method of undetermined coefficients for problem 1. Also, use the ./data/mmdat.zip data to reproduce his figure 3.4. (Pay very careful attention to the notes in the preceding table.) What point is Mark making with this figure?

    Estimated time for completion: 6 hours.

Old EViews Versions of the Assignments

These exercises assume that for every new EViews command you carefully read the online help provided by EViews (see the EViews Command Reference).

  • First Steps: if_hw-eviews.htm#if_intro
  • German Hyperinflation: if_hw-eviews.htm#if_gh
  • Introduction to Unit Roots: macro_hw.htm#ev_uroot
  • Long-Run Purchasing Power Parity: if_hw-eviews.htm#if_ppp
  • Real-Interest-Differential Model: if_hw-eviews.htm#if_rid
  • Spot and Forward Rates: if_hw.eviews.htm#ev_mcc
[acemoglu_etal-2015-handbook]Acemoglu, Daron, et al. (2015) "Democracy, Redistribution, and Inequality". In Anthony B. Atkinson and Fran,cois Bourguignon (Eds.) Handbook of Income Distribution, : Elsevier.
[bumann.lensink-2016-jimf]Bumann, Silke, and Robert Lensink. 2016. Capital Account Liberalization and Income Inequality. Journal of International Money and Finance 61, 143--162.
[candelon.carare.miao-2016-jimf]Candelon, Bertrand, Alina Carare, and Keith Miao. 2016. Revisiting the New Normal Hypothesis. Journal of International Money and Finance 66, 5--31.
[dollar.kraay-2002-jegrowth]Dollar, David, and Aart Kraay. 2002. Growth is Good for the Poor. Journal of Economic Growth 7, 195--225.
[jones.klenow-2016-aer]Jones, Charles I, and Peter J Klenow. 2016. Beyond GDP? Welfare across Countries and Time. American Economic Review 106, 2426--2457.
[kirschenmann.etal-2016-jimf]Kirschenmann, Karolin, Tuomas Malinen, and Henri Nyberg. 2016. The Risk of Financial Crises: Is There a Role for Income Inequality?. Journal of International Money and Finance 68, 161--180.
[ostry.berg.tsangarides-2014-imf]Ostry, Jonathan D, and Andrew Berg. (2014) "Redistribution, Inequality, and Growth". Staff Discussion Note SDN/14/02. https://www.imf.org/external/pubs/ft/sdn/2014/sdn1402.pdf
[stokey-2015-jegrowth]Stokey, Nancy L. 2015. Catching Up and Falling Behind. Journal of Economic Growth 20, .

Mankiw, Romer, and Weil (1992)

Using the data in their appendix, estimate MRW’s equation (7).

Let us extend the Mankiw, Romer and Weil (1992) article. Recall that they implement an empirical test of the neo-classical growth model. For this empirical exercise, first assemble an updated version of MRW’s dataset. Second, attempt to replicate the crosscountry regressions in MRW using your new dataset.

Replication

  1. Create your dataset.
    1. Download the Penn World Table data from the Groningen Growth and Development Centre: https://www.rug.nl/ggdc/productivity/pwt/ Download the Excel version and save the Data sheet as the comma-separated-values file pwt90.csv. (Set the encoding to UTF-8 before saving, using the Tools > Web Options.)
    2. Restrict the sample MRW’s two years: 1960 and 1985.
    3. Which variables from PWT will you use? (Hint: read MRW Data and Samples section I.C to determine which variables do MRW use to measure n, s, and Y/L. Look at the Legend sheet to get the variable names.)
    4. Create the variables you need in order to run the regression estimating the model you derived (Hint: read the note to Table 1 to determine how MRW constructed their variables.)
    5. Create dummy variables for the two rightmost sets of countries MRW use in Table I . (Intermediate and OECD; see lists in MRW.)
  2. Run the first MRW regression using OLS on both subsamples of countries, and report your results.
    1. Are the coefficients as expected?
    2. Do the results support the predictions of the Solow model? Why (not)?
    3. Provide a formal test.
  3. Now we use the fact that the model predicts not just signs of the coeffcients, but also the magnitudes. Rather than estimating the coefficients independently, we can impose a restriction on the regression. Replicate Table 2 in MRW.

Extension

Assembling your dataset

Gather the following variables for as many countries as possible.

  1. Real GDP per worker (PPP-adjusted) for 2000.
  2. Investment share of GDP for the years 1980 and 2000.
  3. (Net) primary and secondary school enrolment rates for 2000.
  4. Population between the ages of 6-11, 12-17 and 15-64 in the years 1980 and 2000.

(1) and (2) can be obtained from the Penn World Tables, version 6.1. (3) can be obtained from the World Development Indicators: HOLLIS has an electronic version of this from which you can download these enrolment rates.

Get (4) from the United Nations website at http://esa.un.org/unpp/: select the “Detailed data” option in the left-hand panel of your webpage to find the relevant variables. Unfortunately, the UN website allows you to download data for only 5 countries at a time. You should therefore only extract the population data for those countries for which the enrolment rate data (from (3)) are also available. Use the “Medium Variant” option when downloading the data.

You will need to match up the data (by country) from the three sources.

Construct the Variables:

  1. Take the average of (I/Y) in 1980 and 2000. This will be your proxy for the capital savings rate, sk.

2. Compute the average annual growth rate of the population aged 15-64 between 1980-2000. This will be your estimate of n.

  1. Assume, as MRW do, that g+ δ = 0.05. Construct the variable n+g+ δ .
  2. Construct a variable called SCHOOL1 = (Net primary school enrolment rate in 2000)*((Population aged 6-11 in 2000)/(Population aged 15-64 in 2000))
  3. Construct an analogous variable called SCHOOL2 = (Net secondary school enrolment rate in 2000)*((Population aged 12-17 in 2000)/(Population aged 15-64 in 2000))

Data Summary and Visualization

  1. Create and discuss appropriate summary tables.
  2. Create and discuss appropriate summary visualizations.

Run MRW regressions

  1. Run a regression based on MRW (7), i.e., the basic Solow model. You may use the full sample of countries that you have. Compare your results to MRW Table 1.
  2. Report the results for both the unrestricted and restricted versions of the regression. Report coefficient estimates, standard errors, R-square, sample size and the implied α values. Derive approximate 95% confidence intervals for those parameters.
  3. Now, test MRW’s “augmented” Solow model. Again, you may use the full sample of countries. Compare your results to MRW Table 2. Run two separate regressions, using SCHOOL1 and SCHOOL2 respectively for the proxy for investment in human capital.
  4. Report the results for both the unrestricted and restricted versions of the regression. Report coefficient estimates, standard errors, R-square, sample size and the implied α and β values. Derive approximate 95% confidence intervals for your parameters.

Write Up Your Results

Address the following:

  1. Introduce the purpose and nature of your empirical analysis.
  2. Provide the estimation equations that you estimate and state what economic theories generated your equations.
  3. Provid a detailed description of your data sources.
  4. Critically discuss the relationship between your theoretical concepts and your empirical constructs. (E.g., how do your data differ from the measures that theory implies?)
  5. Interpret your regression results. (Highlight significant coefficients; discuss fit.)
  6. Are the magnitudes of the implied parameters plausible? Do they support the basic Solow model or the MRW augmented Solow model?
  7. What were the differences when you used SCHOOL1 versus SCHOOL2? Which do you think is the better proxy for investment in human capital? (Explain.)

Extension

DeLong observes that “the production function assumed by MRW has the somewhat odd implication that vastly more of the factor of production human capital is produced when a working-age person from a rich country attends secondary school than when a working-age person from a poor country does.”

Instead, try a Mincerian specification of the role of education and assume the production function: “education amplifies the efficiency of labor beyond its current baseline value Et by a factor equal to the share of the working-age population h attending secondary school raised to the power of the parameter b.”

How does this specification affect your regression results?