Wändi Bruine de Bruin, Charles F. Manski, Giorgio Topa, and Wilbert van der
Klaauw, "Measuring Consumer Uncertainty about Future Inflation", Journal of
Applied Econometrics, Vol. 26, No. 3, 2011, pp. 454-478.
The data used in the paper were collected as part of RAND's American Life
Panel (ALP). For more information about the ALP, see
https://mmicdata.rand.org/alp/index.php/Main_Page
The data were collected through a number of separate survey modules fielded
during the October 2007--October 2009 period. The data we used in our
analysis are organized into two different files, one corresponding to the
"special survey" and one corresponding to the "panel survey", and are
included in both Stata and ASCI format. These datasets only contain the
variables and observations that we actually used in our analyses. Full
descriptions of the questionnaires describing the raw data variables are
provided in PDF format, one for the special survey and one for the final
wave of our panel survey. The raw data files can be downloaded at no cost
(after a simple registration) from RAND's data archive:
https://mmicdata.rand.org/alp/index.php/Data
Also downloadable from this site are the questionnaires and variables
definitions for each survey module. The raw data for the special survey can
be found in the data file corresponding to module 16. The corresponding raw
data files for the panel survey can be found in the data files corresponding
to modules 14, 17, 19, 25, 31, 37, 45, 56, 55, 61, 69, 75, 84 and 89.
As discusses in the paper, ALP participants were divided into (1) an 'old
sample' of individuals aged 40 and older who participated in the Michigan
Survey prior to December 2006, and (2) a 'new sample' of individuals aged 18
and older who participated in the Michigan Survey after December 2006. Those
in the 'old' sample were invited to participate in the panel surveys, and
those in the 'new' sample were invited to participate in the special survey.
A total of 589 participants from the 'new' ALP sample completed our special
survey between December 22, 2007 and May 22, 2008, with 47.9% filling it out
by December 31, 2007, and 86.0% by January 31, 2008.
The first panel survey was fielded on November 7, 2007 and has been repeated
since then every six weeks or so. Here, we report on the first 14 waves with
the most recent one entering the field on July 31, 2009. In our analysis of
each panel survey, we only consider responses for those participants who fill
it out within 30 days after the field date, in order to avoid spurious
heterogeneity in responses due to changing economic conditions over time. In
addition, our analyses include only those respondents who participated in at
least five of the first nine waves. These criteria yield a panel with fairly
stable composition and number of responses over time, with on average about
400 responses per survey.
In some of the modules that are part of the panel survey, we randomized
between respondents, where one group (group A) would receive questions Q20
and QExtra, while a second group (group B) was asked questions Q20B and
QExtraB. The difference between the questions was that group A was only
shown 8 bins to allocate probabilities over, while group B was shown 10 bins
instead. In our analysis, we only used answers provided to questions Q20 and
QExtra, as we have a longer time series for these questions.
Accompanying files
Msall-JAE.dta Stata datafile containing panelsurvey data
Msall-JAE.txt ASCI datafile containing panelsurvey data
Specialsurvey-JAE.dta Stata datafile containing specialsurvey data
Specialsurvey-JAE.txt ASCI datafile containing specialsurvey data
Panelsurvey-codebook codebook for last module of panel survey (ms89)
Specialsurvey-codebook codebook of special survey (ms16)
Panelsurvey-quest questionnaire of last module of panel survey (ms89)
Specialsurvey-quest questionnaire of special survey (ms16)
The descriptions below list the variables in the order in which they appear
in the datasets.
Description of Variables in Panel Survey (msall-JAE)
Prim_key: Individual identifier to link records belonging to the
same individual over time
Tsstart: Time and date when started survey
Tsend: Time and date completed survey
Tsend_formatted, Tsend_formatted_td: formatted versions of Tsend
Procrastinate: Number of days since module was fielded (in our analysis,
we excluded responses more than one month after the date
the survey was fielded).
Module_id: Survey module number (wave of panel)
randomVersion: flag variable for whether individuals were given 8-bin or
10-bin density questions. randomVersion==1 means 8-bin version,
randomVersion==2 means 10-bin version.
Q8, Q9, Q10, Q10_rg1, Q10_rg2, Q11_1, Q11_2, Q11_r, Q11_f, Q12, Q12_f,
FLQ12, Q13, Q14, Q14_f, Q15, Q15_rg1, Q15_rg2, Q16_1, Q16_2, Q16_r, Q16_f,
Q17, Q17_f, FLQ17, Q18, Q19, Q19_f: See enclosed codebook for panel survey
and questionnaire for MS89 (latest module included in panel) for
descriptions of these variables measuring respondents' point forecast of
year-ahead changes in prices in general.
Infl_mi Respondent's point forecast of year-ahead changes in prices in
general (computed using the variables Q8 to Q19_f listed above)
Q20Intro, Q20_a, Q20_b, Q20_c, Q20_d, Q20_e, Q20_f, Q20_g, Q20_h, Q20Total:
See enclosed codebook for panel survey and questionnaire for MS89 (latest
module included in panel) for descriptions of these variables measuring
respondents' percent chance associated with different intervals of
year-ahead changes in prices in general 8-bin version.
Q20BIntro, Q20B_a, Q20B_b, Q20B_c, Q20B_d, Q20B_e, Q20B_f, Q20B_g, Q20B_h,
Q20B_i, Q20B_j, Q20BTotal: same as above but now 10-bin version.
Mean_mi: Estimated mean of generalized Beta distribution for changes in
prices in general
Median_mi: Estimated median of generalized Beta distribution for changes
in prices in general
Var_mi: Estimated variance of generalized Beta distribution changes for
in prices in general
Quant1_mi: Estimated 25th percentile of generalized Beta distribution for
prices in general
Quant3_mi: Estimated 75th percentile of generalized Beta distribution for
prices in general
Q37, Q38, Q38_rg1, Q38_rg2, Q38_r, Q39, Q39_rg1, Q39_rg2, Q39_r: See
enclosed codebook for panel survey and questionnaire for MS89 (latest module
included in panel) for descriptions of these variables measuring
respondents' point forecast of year-ahead changes in earnings.
Wage Respondent's point forecast of year-ahead changes in earnings
(computed using the variables Q37 to Q39_r listed above)
QExtra_a, QExtra_b, QExtra_c, QExtra_d, QExtra_e, QExtra_f, QExtra_g,
QExtra_h, QExtraTotal: See enclosed codebook for panel survey and
questionnaire for MS89 (latest module included in panel) for descriptions of
these variables measuring respondents' percent chance associated with
different intervals of year-ahead growth in earnings 8-bin version (only
asked to those working).
QExtraB_a, QExtraB_b, QExtraB_c, QExtraB_d, QExtraB_e, QExtraB_f, QExtraB_g,
QExtraB_h, QExtraB_i, QExtraB_j, QExtraBTotal: same as above but now 10-bin
version.
Mean_wg: Estimated mean of generalized Beta distribution for changes
in wage earnings
Median_wg: Estimated median of generalized Beta distribution for changes
in wage earnings
Var_wg: Estimated variance of generalized Beta distribution for
wage earnings
Quant1_wg: Estimated 25th percentile of generalized Beta distribution for
wage earnings
Quant3_wg: Estimated 75th percentile of generalized Beta distribution for
wage earnings
Birthday: day of birth
Birthmonth: month of birth
Birthyear: year of birth
Calcage: computed age (in years) of respondent at time of survey
Ageover59: Indicator for whether 60 or older at time of survey
Highesteducation: highest degree attained
College: Indicator for whether respondent has BA or higher degree
Currentlivingsituation: (1) Married or living with a partner, (2) Separated,
(3) Divorced, (4) Widowed, (5) Never Married
Married: Indicator for whether married/living together or not
Gender: (1) Male, (2) Female
Male: Indicator for whether respondent is male
Race: (1) white/Caucasian, (2) black/African American, (3) American
Indian/Alaskan, (4) Asian/Pacific Islander, (5) other
Familyincome: (1) Less than 5K, (2) 5K to 7.5K, (3) 7.5K to 10K, (4)10K to
12.5K, (5) 12.5K to 15K, (6) 15K to 20K, (7) 20K to 25K,
(8) 25K to 30K, (9) 30K to 35K, (10) 35K to 40K, (11) 40K to 50K,
(12) 50K to 60K, (13) 60K to 75K, (14) 75K or more.
Familyincome_part2: (1) 75K to 100K, (2) 100K to 125K, (3) 125K to 200K,
(4) 200K or more
Incover40: Indicator for whether family income is at least 40K
Incover75: Indicator for whether family income is at least 75K
Q31s1: indicator for whether currently working
Numeracy: Total number of correct answers on numeracy questions
Finlit: Indicator for correct answer to single financial literacy
question
Highfinlit: Indicator for high numeracy&financial literacy (combined) score
Description of Variables in Special Survey (specialsurvey-JAE)
Prim_key: Individual identifier to link records belonging to the same
individual over time
Tsstart: Time and date when started survey
Tsend: Time and date completed survey
Q8, Q9, QA9, Q10, Q11_1, Q11_2, Q11_r, Q11_f, Q12, FLQ12, Q13, Q14, Q15,
Q16_1, Q16_2, Q16_r, Q16_f, Q17, FLQ17, Q18, Q19: See enclosed codebook and
questionnaire of special survey for descriptions of these variables
measuring respondents' point forecast of year-ahead changes in prices in
general.
Infl_mi: Respondent's point forecast of year-ahead changes in prices
in general (computed using the variables Q8 to Q19 listed above)
P012: How hard to answer question about prices in general -- scale
1 (very easy) to 7 (very hard)
P013: How clear was question about prices in general -- scale 1
(very unclear) to 7 (very clear)
P022a, P022b, P022c, P022d, P022e, P022f, P022g, P022h, P022Total: See
enclosed codebook and questionnaire of special survey for descriptions of
these variables measuring respondents' percent chance associated with
different intervals of year-ahead changes in prices in general with 8 bins.
P023: How hard was it to answer probabilistic question about prices
in general -- scale 1 (very easy) to 7 (very hard)
P024: How hard was it to give answers that added to 100 -- scale 1
(very easy) to 7 (very hard)
P025: How clear was probabilistic question about prices in general --
scale 1 (very unclear) to 7 (very clear)
Mean_mi: Estimated mean of generalized Beta distribution for changes
in prices in general
Median_mi: Estimated median of generalized Beta distribution for changes
in prices in general
Var_mi: Estimated variance of generalized Beta distribution for changes
in prices in general
Quant1_mi: Estimated 25th percentile of generalized Beta distribution for
prices in general
Quant3_mi: Estimated 75th percentile of generalized Beta distribution for
prices in general
WR008,WR009, WR010_1, WR010_2, WR010_r: See enclosed codebook and
questionnaire of special survey for descriptions of these variables
measuring respondents' point forecast of year-ahead changes in earnings.
Wage Respondent's point forecast of year-ahead changes in earnings
(computed using the variables WR008 to WR010_r listed above)
WR013: How hard to answer question about earnings -- scale 1
(very easy) to 7 (very hard)
WR014: How clear was question about earnings -- scale 1 (very unclear)
to 7 (very clear)
WR015_a, WR015_b, WR015_c, WR015_d, WR015_e, WR015_f, WR015_g, WR015_h,
WR015Total: See enclosed codebook and questionnaire of special survey for
descriptions of these variables measuring respondents' percent chance
associated with different intervals of year-ahead growth in earnings (only
asked to those working).
Mean_ny: Estimated mean of generalized Beta distribution for changes in
wage earnings
Median_ny: Estimated median of generalized Beta distribution for changes
in wage earnings
Var_ny: Estimated variance of generalized Beta distribution for wage
earnings
Quant1_ny: Estimated 25th percentile of generalized Beta distribution for
wage earnings
Quant3_ny: Estimated 75th percentile of generalized Beta distribution for
wage earnings
FD002d: Indicator for respondent taking main responsibility for
investing in household
FD003: Planning Horizon Spending from (1) next day, to (9) longer
than 10 years
FD004: Planning Horizon Saving from (1) next day, to (9) longer than
10 years
Birthday: day of birth
Birthmonth: month of birth
Birthyear: year of birth
Calcage: computed age (in years) of respondent at time of survey
Highesteducation: highest degree attained (see codebook)
Currentlivingsituation: (1) Married or living with a partner, (2) Separated,
(3) Divorced, (4) Widowed, (5) Never Married
Gender: (1) Male, (2) Female
White: (1) white, (2) black, (3) native American, (4) Asian, (5) other
Familyincome: (1) Less than 5K, (2) 5K to 7.5K, (3) 7.5K to 10K, (4) 10K
to 12.5K, (5) 12.5K to 15K, (6) 15K to 20K, (7) 20K to 25K,
(8) 25K to 30K, (9) 30K to 35K, (10) 35K to 40K, (11) 40K to 50K,
(12) 50K to 60K, (13) 60K to 75K, (14) 75K or more.
WR001s1: current employment status = working now
Numeracy: Total number of correct answers on numeracy questions
Finlit: Indicator for correct answer to single financial literacy
question
Finlit_total: Total financial literacy score (Finlit and numeracy combined))
There are four files, two of which are zip files that contain six other
files in total.
bmtk-data-txt.zip contains:
specialsurvey-JAE.txt
specialsurvey-codebook.txt
panelsurvey-codebook.txt
msall-JAE.txt
bmtk-data-dta.zip contains:
specialsurvey-JAE.dta
msall-JAE.dta
specialsurvey-quest.pdf
panelsurvey-quest.pdf
Unix/Linux users should use "unzip -a" when unzipping bmtk-data-txt.zip but
*not* when unzipping bmtk-data-dta.zip.