Proofreading #250

paper_slides/input/test_stateFE.tex

@@ -26,7 +26,8 @@
         The third column shows the value of the difference of the relevant coefficients,
         and the fourth column shows the standard error of the difference.
         The fifth and sixth columns show the implied $t$- and $p$-values.
-        All regressions include economic controls from the QCEW.
+        We constructed these tests by estimating both regression equations jointly, 
+        including economic controls from the QCEW as in the paper.
         Standard errors in parentheses are clustered at the state level.
     \end{minipage}
 \end{table}

paper_slides/reply/cover.lyx

@@ -234,12 +234,12 @@ Our main results are obtained using a monthly panel dataset.
  The editor questioned the plausibility of these dynamics.
  The argument is that the effect should be sluggish as existing rental contracts
  are re-negotiated over time.
- Referee 3 (R3) suggests that we these patterns arise because our measure
- of rents reflects posted prices in newly-available rental units, rather
- than rates of existing contracts.
+ Referee 3 (R3) suggests that these patterns arise because our measure of
+ rents reflects posted prices in newly-available rental units, rather than
+ rates of existing contracts.
  We agree.
- Furthermore, we see this as a feature of our rents measure, as rents of
- new contracts are more reflective of market conditions (Ambrose et al.
+ Furthermore, we see this as a feature of our rents measure, since rents
+ of new contracts are more reflective of market conditions (Ambrose et al.
 \begin_inset ERT
 status open
 
@@ -253,7 +253,6 @@ status open
 \end_inset
 
  2015).
-
 \begin_inset Foot
 status open
 
@@ -280,7 +279,16 @@ literal "false"
 \end_inset
 
  to see our question.
- (SH: Is this useful?)
+\begin_inset Note Note
+status open
+
+\begin_layout Plain Layout
+SH: Is this useful?
+\end_layout
+
+\end_inset
+
+
 \end_layout
 
 \end_inset
@@ -290,11 +298,11 @@ literal "false"
 \end_layout
 
 \begin_layout Itemize
-When using a yearly model and our baseline sample of ZIP codes we find no
- significant results (Online Appendix Table 3, Panel C), which the AER editor
- and R1 found troublesome.
- This model is simply an averaged version of the monthly model, so it tries
- to get at the same coefficients.
+When using a yearly model and our baseline sample of ZIP codes we find similar
+ patterns in point estimates but no statistical significance (Online Appendix
+ Table 3, Panel C), which the AER editor and R1 found troublesome.
+ The yearly model is simply an averaged version of the monthly model, so
+ it tries to get at the same coefficients.
 \begin_inset Foot
 status open
 
@@ -336,15 +344,14 @@ Omitting the controls for simplicity, the monthly model can be written as
 
 \end_inset
 
- We think that the reason for this is lack of power.
+ However, statistical power is much different across these models.
  In fact, comparing rows (iii) across Panels A and C of Online Appendix
- Table 3, we find that standard errors are between 3.8 and 4.6 times in the
- yearly model (note that the clustering scheme is constant).
- As a result, we find that this model does not reject our baseline estimates.
+ Table 3 for our rents outcome, we find that state-clustered standard errors
+ are between 3.8 and 4.6 times larger in the yearly model.
+ As a result, the yearly model does not reject our baseline estimates.
  There are two reasons for this.
- First, the number of observations: the yearly model is estimated on 1/12th
- the number of observations.
- Second, as we mention in the paper, the smoothing of identifying variation
+ First, the yearly model is estimated on 1/12th the number of observations.
+ Second, as we mention in the paper, the yearly averages smooth useful variation
  when we average MW changes that happened in the middle of the year (most
  commonly, July).
  We included these estimates to illustrate the importance of using monthly
@@ -442,9 +449,9 @@ not
  level is not available.
  In the paper we cite related literature that explores this channel (e.g.,
  Leung 2021).
- We included this variable is to account for spatial heterogeneity which
- both our model and the empirical results suggest is important, especially
- in the heterogeneity analyses in Table 5.
+ We included this variable to account for spatial heterogeneity which both
+ our model and the empirical results suggest is important, especially in
+ the heterogeneity analyses in Table 5.
  We will revise the paper to note that due to data constraints we cannot
  test this channel directly and that more work is needed to conclusively
  establish that the negative coefficient arises from changes in local prices.
@@ -675,42 +682,52 @@ median
  We will add this discussion to the paper and include a new robustness analysis
  that uses a newly available variable in Zillow that directly attempts to
  control for selection of listings.
- We provide more details on this new variable below.
+ We provide more details on this new variable below, where we discuss alternativ
+e sources of data more generally.
 \end_layout
 
 \end_deeper
 \begin_layout Enumerate
 We also received comments regarding our estimates of the effect of the MW
  on income.
- These estimates are discussed in Appendix D and displayed in Online Appendix
- Table 7, and show that a 10% increase in the workplace MW leads to a roughly
+ These estimates, discussed in Appendix D and displayed in Online Appendix
+ Table 7, show that a 10% increase in the workplace MW leads to a roughly
  1% increase in wage income in a ZIP code.
  Before going over the comments, we note that the goal of the paper is 
 \emph on
 not
 \emph default
- to estimate the effect of the MW on wage income, but rather to set a sensible
- value for the parameter 
+ to estimate the elasticity of wage income to the MW (
+\begin_inset Formula $\varepsilon$
+\end_inset
+
+), but rather to select a sensible value for 
 \begin_inset Formula $\varepsilon$
 \end_inset
 
  in Section 6.
- Nonetheless, we understand that an inacurrate value for 
+ This is the reason our analysis in Appendix D is brief.
+ Nonetheless, we understand that an inacurrate choice for 
 \begin_inset Formula $\varepsilon$
 \end_inset
 
  would lead to incorrect conclusions in the counterfactual exercises.
- We will address these comments expanding Section 6 by adding a figure that
- shows how the conclusions of the counterfactuals change with different
- values of 
+ We will expand Section 6 by adding a figure that shows how the conclusions
+ of the counterfactuals change with different values of the elasticity 
 \begin_inset Formula $\varepsilon$
 \end_inset
 
 .
- As for Appendix D, we can justify our parameter value for the elasticity
- of income to the MW from the literature, so we are willing to drop it entirely.
- If you believe that providing valid estimates of this parameter is an important
- contribution, we could either maintain Appendix D with minor additions
+ As for Appendix D, we will justify our choice of 
+\begin_inset Formula $\varepsilon$
+\end_inset
+
+ from the literature, so we are willing to drop it entirely.
+ If you believe that providing valid estimates of 
+\begin_inset Formula $\varepsilon$
+\end_inset
+
+ is an important contribution, we could either maintain it with minor additions
  or deep dive in the analysis and move it into the main paper.
 \end_layout
 
@@ -837,7 +854,8 @@ status open
 
 \end_inset
 
- (2019) find spillovers up to $3 above the new MW.
+ (2019) find spillovers up to $3 above the new MW, and estimate that they
+ account for 40% of the wage effect of the MW.
  Aaronson and French (2007, Table 1) estimates that the share of the wage
  bill affected by the MW in the restaurant industry is 0.17 (the share of
  MW workers in the industry is 0.33).
@@ -911,7 +929,7 @@ status open
 
 \begin_layout Plain Layout
 We emphasize that, as long as workers can in principle move, prices will
- respond even if they decide to stay in the same location.
+ respond even if they decide to stay in the same housing unit.
 \end_layout
 
 \end_inset
@@ -920,7 +938,7 @@ We emphasize that, as long as workers can in principle move, prices will
  outside their own ZIP code as a response to the MW.
  In that case we would end up with an extra term for the changes in commuting
  shares in Proposition 2.
- This term would go to the residual in our main models.
+ This term would go to the residual in our empirical estimates.
  We observe changes in commuting shares annually, and use time-varying commuting
  shares to construct our workplace MW measure in Panel B of Table 3, finding
  similar results.
@@ -973,12 +991,13 @@ shopping MW
 \begin_layout Standard
 We now discuss alternative sources of data.
  As we discuss in the introduction of the paper, a key challenge to study
- the effect of the minimum wage on rents is the existence high quality data.
+ the effect of the minimum wage on rents within-cities is the existence
+ of high quality data.
  With its limitations, we see our estimates using Zillow data as quantiatively
  reasonable and providing a novel contribution to the literature.
- However, we acknoweledge that using alternative sources would further increase
- confidence in our results.
- Thus, we plan to explore new data sources that have became available since
+ However, we acknoweledge that using alternative data sources would further
+ increase confidence in our results.
+ Thus, we plan to explore two data sources that have became available since
  the first version of our paper.
 
 \end_layout
@@ -988,7 +1007,7 @@ We now discuss alternative sources of data.
 We will add estimates using Zillow's new monthly rental housing index (ZORI).
  This index, based on methodology by Ambrose et al.
  (2015), directly attempts to control for changes in the composition of
- posted rents on the platform by measuring the change in rents for the same
+ posted rents on the platform by estimating change in rents for the same
  rental unit over time.
 \begin_inset Note Note
 status open
@@ -1018,18 +1037,18 @@ literal "false"
 We will pull ZIP code-level monthly data on rents from a newly available
  API by Realty Mole.
  These data provide the average rent and also incorporates the number of
- units with a different number of bedrooms that make up the average, potentially
- allowing us to study for changes in composition of rentals.
+ units with a different number of bedrooms that make up that average, potentiall
+y allowing us to study changes in the composition of rentals.
  The downside of these data is that it starts in mid 2020.
  If we find these data to be of good quality we might introduce new robustness
  analysis in the paper.
 \begin_inset Note Note
 status open
 
 \begin_layout Plain Layout
-SH: We could also mention the county-level data from the FMR to have a more
- robust statistic of the yearly distribution of rents, addressing AER editor
- comment.
+SH: We could also mention the county-level data from FMR to have a more
+ robust statistic of the yearly distribution of rents, addressing one of
+ the comments by the AER editor.
 \end_layout
 
 \end_inset
@@ -1115,7 +1134,7 @@ not occupied by poor workers,
  This claim comes from Online Appendix Table 6, where we estimate our model
  in different housing categories in the Zillow data.
  We discuss those results in Section 5.4, where we warn against strong conclusion
-s as these estimates are noisy.
+s as these estimates are quite noisy.
  We note that these rental categories are inhabited by low-income households
  as well (Online Appendix Figure 3), though we acknowledge that the share
  of low-wage households in some categories is relatively lower.
@@ -1167,6 +1186,30 @@ clearpage
 \end_inset
 
 
+\end_layout
+
+\begin_layout Standard
+\align center
+\begin_inset CommandInset include
+LatexCommand input
+filename "../input/test_stateFE.tex"
+
+\end_inset
+
+
+\begin_inset ERT
+status open
+
+\begin_layout Plain Layout
+
+
+\backslash
+vspace{3mm}
+\end_layout
+
+\end_inset
+
+
 \end_layout
 
 \begin_layout Subsection*
@@ -1357,26 +1400,6 @@ makeatother
 \end_inset
 
 
-\begin_inset ERT
-status open
-
-\begin_layout Plain Layout
-
-
-\backslash
-vspace{3mm}
-\end_layout
-
-\end_inset
-
-
-\begin_inset CommandInset include
-LatexCommand input
-filename "../input/test_stateFE.tex"
-
-\end_inset
-
-
 \end_layout
 
 \end_body