Table 2.

Changes in noodle purchases in response to the Scandal (in grams/month).

	Unweighted			Inverse propensity weighting
Buyer group	Monthly buyer	Frequent buyer	Infrequent buyer	Monthly buyer	Frequent buyer	Infrequent buyer
Comparison group	Non-buyer			Non-buyer
	(1)	(2)	(3)	(4)	(5)	(6)
Panel A: Non Maggi Noodles (Reference period: April 2015)
Immediate response	62.396***	18.861***	−8.058***	64.001***	22.720***	−8.932***
(May 2015–Nov 2015)	(5.397)	(3.468)	(3.237)	(5.595)	(9.356)	(3.768)
1-year post response	72.969***	36.672***	2.933	66.660***	36.130***	2.919
(Dec 2015–Nov 2016)	(5.580)	(3.378)	(3.147)	(5.561)	(6.794)	(3.153)
2-year post response	35.110***	14.466***	−4.822	31.853***	15.625	−4.418
(Dec 2016 -Nov 2017)	(5.397)	(3.577)	(3.371)	(5.729)	(9.143)	(3.634)
Panel B: All noodles (Reference period: April 2015)
Immediate response	–355.051***	–214.231***	–68.169***	–361.683***	–214.541***	–70.180***
(May 2015–Nov 2015)	(12.739)	(4.146)	(3.387)	(13.374)	(9.653)	(3.949)
1-year post response	–148.871***	–111.824***	–42.385***	–143.852***	–115.186***	–43.188***
(Dec 2015–Nov 2016)	(12.075)	(4.060)	(3.342)	(12.159)	(6.379)	(3.269)
2-year post response	–119.481***	–85.221***	–36.703***	–117.681***	–95.579***	–38.102***
(Dec 2016–Nov 2017)	(12.957)	(4.417)	(3.699)	(14.364)	(12.822)	(3.945)
Household fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
N	191,160	1,137,672	1,420,848	191,160	1,137,672	1,420,848

Note: Columns 4–6 present the average event study estimates within the corresponding period shown in Figs. 3 and 4. Buyer groups are defined based on the number of months they purchased Maggi noodles during the 24 months prior to May 2015: 24 months for Monthly buyer; 13–23 months for Frequent buyer; 1–12 months for Infrequent buyer. Comparsion households are those who did not buy any Maggi noodles for 2 years prior to May 2015. Robust standard errors clustered at household level.

^* P < 0.1.

^** P < 0.05.

^*** P < 0.01.