eco question

Economics 5140 �American Economic History

Prof. T.D. Logan, Spring Semester 2018

Midterm Exam 1, Part 1 February 7, 2018

Answer all parts of all questions in your bluebook. Be precise and concise in your answers, and make sure that you read the entire exam before you begin. Relax, take a deep

breath, and do your best. Good Luck!!

1

1 Macroeconomic Measurement in Economic History

1. (15) Balke & Gordon ("The Estimation of Prewar Gross National Product") use both a com- ponents and indicators method to estimate GNP. Explain one way in which their indicator method incolumn(1) (producedabove) isdi¤erent fromandoneway inwhich it is similar to the indicator method of Romer ("The Prewar Business Cycle Reconsidered") [be speci�c]? Does it yield similar results to Romer�s indicator method? Why or why not [justify your answer]? Column (3) is analogous to the Kuznets indicator method as described by Romer. Describe in detail one reason why Romer argues that the indicator speci�cation in column (3) may not be the appropriate way to derive estimates of GNP.

2. (10) Romer ("Spurious Volatility in Historical Unemployment Data") adopts a novel approach todiscoveringwhetherunemploymenthasdecreased involatilityover time. DescribeRomeriza- tion, both in abstract as a general methodological approach and, in particular, how Romer used that method is used to answer the question of whether the unemployment rate has decreased in cyclical volatility over time.

3. (10) Describe three ways in which Davis�series for Industrial Production is di¤erent than the existing series. (Be as speci�c as you can. For example, what time span does it cover? Does it contain more or fewer government sources? Does it contain more or fewer actual output estimates?) Using the index, does Davis �nd that business cycle volatility changed before and after the Civil War? What measure does he use to compare the pre-Civil War and post-Civil War business cycles? [Hint: this is a di¤erent issue than the volatility question considered by Romer.]

4. (15)Consider theestimationofCPIbiasusingEngel curvesbyCosta. Recall that theestimating equation is wi;j;t = �+'(lnPF;j;t � lnPN;j;t)+� (lnYi;j;t � lnPj;t)+X�+ui;j;t where w is the share of the budget devoted to food for household i in year t in place j, P is the true, unobserved

2

price index for food (F), nonfood (N), and all goods, Y is household expenditure, X is a vector of household characteristics, and u is the error term. Assuming that prices are measured with error such that ln(Pj;t) = ln(Pj;0) + ln(1+�j;t) + ln(1+Ej;t); ln(PF;j;t) = ln(PF;j;0) + ln(1+�F;j;t)+ln(1+EF;j;t); and ln(PN;j;t) = ln(PN;j;0)+ln(1+�N;j;t)+ln(1+EN;j;t) where P0 is the true price at time 0, � is the CPI, and E is the cumulative measurement error in the price index from year 0 to year t. Name one reason why food is chosen for the Engel curve analysis (as opposed to another type of expenditure) and one assumption needed for this model to be used to derive estimates of CPI bias.

Recall that the functional form of estimating CPI bias

wi;j;t =

� +' [ln(1+�F;j;t)� ln(1+�N;j;t)]+� [lnYi;j;t � ln(1+�j;t)]+X� + TX t=1

�tDt + X j=1

�jDj +ui;j;t

where Dj are regional e¤ects and time e¤ects from year 0 to year t are captured in Dt, follows from the above. Show that the time drift in the Engel curve estimates is equal to CPI bias, that is, show that �t = ' [ln(1+EF;j;t)� ln(1+EN;j;t)] � � ln(1+Ej;t) [show all of your work, partial credit will be awarded]. What omitted time-varying factor does Logan ("Are Engel Curve Estimates of CPI Bias Biased?") claim is erroneously included in Costa�s estimates of �t? [Note: you do NOT need to write down any equations for this last question.]

3