Professional Documents
Culture Documents
SupplementaryNotes FE
SupplementaryNotes FE
• • • • • • • • • • •
-5 -4 -3 -2 -1 0 1 2 3 4 5
t
Case II: Temperature remains at T̄it ∀ t 6= 0. But there is a negative temperature shock at
period t = 0, i.e., Ti0 < T̄it = 0. The dynamic effects of this temperature shock on output
depend on the underlying parameters β and γ. To see this, let’s consider two sub-cases.
1
following figure depicts the time series of yit . In this case, we say temperature only has level
effect (temporary effect) on economic growth but no growth effect (permanent effect).
• • • • • • • • • •
-5 -4 -3 -2 -1 0 1 2 3 4 5
t
∆Ai,0
= γTi,0 > 0 when t = 0
Ai,−1
Hence, the productivity Ait follows the path ... = Ai,−2 = Ai,−1 < Ai,0 = Ai,1 = Ai,2 = ....
Therefore, output also follows the path ... = yi,−2 = yi,−1 < yi,0 = yi,1 = yi,2 = .... In
this case, we say temperature has growth effect (permanent effect) on economic growth. The
following figure depicts the time series of yit . Obviously, the effect temperature shock persists
forever.
• • • • • •
temporary temperature shock will have a permenent effect on productitivity!
• • • • •
-5 -4 -3 -2 -1 0 1 2 3 4 5
t
scar forever!
2
2 An Extended Model
Consider the following simple economy:
∆Ait
= gi + γ0 Tit + γ1 Tit−1 (2)
Ait−1
where Y is the aggregate output, L measures population, A measures labor productivity
and T measures weather. Equation (1) captures the level effect of weather on production,
e.g., the effect of current temperature (Tit ) and lag period’s temperature (Tit−1 ) on crop
yields. Equation (2) captures the growth effect of weather, e.g., the effect of current and
past temperature on features, such as institutions that influence productivity growth. gi
captures all the time-invariant country-specific factors that affect productivity growth.
PG PG
Denote G be the number of periods. Define T̄it = t=0 Tit /G, T̄it−1 = t=0 Tit−1 /G,
PG
T̄it−2 = t=0 Tit−2 /G. In the following analysis, we assume that when G is sufficiently large,
T̄it = T̄it−1 = T̄it−2
(1) Denote output per capital as yit = Yit /Lit and economic growth rate as git = ∆yit /yit−1 .
Show that the economic growth rate at period t is:
3
∆yit ∆yit ∆Ait ∆Ait
Using the fact that ln(1 + yit
) ≈ yit
and ln(1 + Ait
) ≈ Ait
, we can rewrite the
above equation as
PG
(2) Show that the fundamental growth rate of the ḡit = t=0 git /G is:
Explain intuitively why level effects β0 and β1 disappear from equation (4). What is
the growth effect of temperature?
Answer: Take average across time on both sides of equation (3) and obtain
Using the fact that T̄it = T̄it−1 = T̄it−2 , the above equation will boil down to equation
(4).
(3) Suppose that you have a panel data that covers I number of countries and G period
of time. Write down a regression model based on the theoretical model (3).
4
where θi , ρ0 , ρ1 and ρ2 correspond to gi , (β0 + γ0 ), (−β0 + β1 + γ1 ) and −β1 in equation
(3).
(3) Discuss how will you test the null hypotheses: (i) The immediate effect of temperature
on economic growth is zero; (ii) The growth effect of temperature on economic growth
is zero.
Answer: Based on the regression model (5), we can test the following hypotheses:
(i) H0 : ρ0 = 0 (The immediate effect of temperature on economic growth is zero)
(ii) H0 : ρ0 + ρ1 + ρ2 = 0 (The growth effect of temperature on economic growth is
zero)