All Tests that is in the Version 2.0 of the article
#Research Funding (USD) #Total_Amount_Grants = Total amount of all
other grants besides: ‘I-Corps’, ‘I-Corps Hubs’, ‘I-Corps-Sites’,
‘I-Corps-Nodes’,‘SBIR_1’, ‘SBIR_2’,‘STTR_1’,‘STTR_2’ # Log Research
Funding (USD) #log_Total_Amount_Grants = just log of Total_Amount_Grants
+1 to avoid problems # Research (annual count) # Total_Grants = Sum
count of all unique research happening in the year (meaning grants) #
Log Research (annual count) #log_Total_Grants = LOG of Total_Grants #
Program Grants (USD) #Amount_ECO = Amount of all I-Corps grants #
Program Grant (annual) #d_ECO = 0 when no Amount_ECO, 1 only for the
year/city receiving Amount_ECO # Program Grant (post) #d_f_ECO = 0 until
get Amount_ECO, 1 after that and stays 1 forever # University (annual
count) #Numb_active_research = Unique Institutions in the current year
doing R&D (meaning they got some money) # Ecosystem Grants (annual)
#d_HNS = 0 when no Amount_HubsNodesSites, 1 only for the year/city
receiving Amount_HubsNodesSites # Ecosystem Grants (post) #d_f_HNS = 0
until get Amount_HubsNodesSites, 1 after that and stays 1 forever #
Investigators (annual count) #Total_Investigator = Total (Sum) number of
investigators by year by city (this is the sum of Count_Investigator1,
2, 3) # Log Private Funding (USD) #log_funding_total = Total funding CB
ANY type of funding (Angel, VC, PE and so on)
# Private Funding (count) #funding_count = Number of funding in the city
# Companies Founded (annual count) #companies_founded = sum count of all
companies that started in the current year
# Training Grants (USD) #log_Amount_I_Corps = I-Corps Teams – ####
Columns not used
# dc_f_Amount_ECO = 0 for cities not treated, 1 for treated (this is
separating all the cities never treated against cities treated, which
trated meaning I-Corps) # d_2011 = 0 for non treated cities and 0 for
cities Amount_ECO (treated) for the years prior to 2011, 1 after 2011. #
dA_2011 = 0 for all Cities prior 2011, 1 after 2011 (Just give a dummy
for 2011) # Count_Investigator1 = Sum count of all unique research
happening in the year (meaning grants) #This is the number of the
leading investigators this is also the number o UNIQUE grants, so you
might have cases where the same investigator will have 2 or more active
grants happening at the same time, so this will count the total (the
investigator will appear twice in this example) Count_Investigator2 and
Count_Investigator3 are the same case, but you should ignore those
columns, instead use the next column # Count_I-Corps = count of I-Corps
teams – # log_population = LOG Population equal before 2018 and with
current years after that. Most cities are undefined # Unique_NSFID1=
This is the unique number of leading investigators, different from the
Count_Investigator1, it will exclude duplicates, this can be useful if
we need to know if the leading investigator are leading more than one
grant at the same time (just by subtracting this from the
Count_Investigator1) # I am also including all the columns from the
University datasets that were “convertible” to city level. For example,
R1 and R2 are not convertible so I am not keeping, but we could “count”
that if you want, just be aware that any university level data is only
available for a few places so most likely will not be very helpful
anyway. # unique_companies_count = number of companies receiving
investments by year by city. This is calculated out of the $ amount from
CB –
####0
feglm_0 = feglm(companies_founded ~ log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = diddata, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,836/0 fixed-effects (29,376 observations) removed because of only 0 outcomes or singletons.summary(feglm_0)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 14,656
## Fixed-effects: location_id: 916, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## log_Total_Amount_Grants -0.001989 0.004231 -0.470038 6.3834e-01
## log_funding_total 0.038148 0.004771 7.995798 1.3898e-15 ***
## funding_count 0.000316 0.000320 0.987468 3.2343e-01
## Total_Grants 0.002653 0.000734 3.614042 3.0254e-04 ***
## unique_companies_count -0.000103 0.000244 -0.421868 6.7313e-01
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.940505######## feglm_h1
feglm_h1 = feglm(companies_founded ~ d_ECO + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = diddata, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,836/0 fixed-effects (29,376 observations) removed because of only 0 outcomes or singletons.summary(feglm_h1)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 14,656
## Fixed-effects: location_id: 916, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 0.398119 0.027235 14.618115 < 2.2e-16 ***
## log_Total_Amount_Grants -0.003562 0.004156 -0.857125 3.9139e-01
## log_funding_total 0.037442 0.004684 7.993645 1.4142e-15 ***
## funding_count 0.000704 0.000314 2.241593 2.5004e-02 *
## Total_Grants 0.002601 0.000718 3.620809 2.9474e-04 ***
## unique_companies_count -0.000538 0.000241 -2.230514 2.5729e-02 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.933401######## feglm_h2 & feglm_h3 - moderations
#MAIN
feglm_h2 = feglm(companies_founded ~ d_ECO * log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = diddata, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,836/0 fixed-effects (29,376 observations) removed because of only 0 outcomes or singletons.summary(feglm_h2)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 14,656
## Fixed-effects: location_id: 916, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 2.743211 0.256543 10.693004 < 2.2e-16 ***
## log_Total_Amount_Grants -0.000986 0.004163 -0.236827 8.1279e-01
## log_funding_total 0.036008 0.004694 7.671614 1.8126e-14 ***
## funding_count 0.000588 0.000312 1.884170 5.9563e-02 .
## Total_Grants 0.002423 0.000716 3.383635 7.1733e-04 ***
## unique_companies_count -0.000381 0.000240 -1.585100 1.1297e-01
## d_ECO:log_Total_Amount_Grants -0.137428 0.014965 -9.183345 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.932205#MAIN
feglm_h3 = feglm(companies_founded ~ d_ECO * log_funding_total + log_Total_Amount_Grants +
funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = diddata, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,836/0 fixed-effects (29,376 observations) removed because of only 0 outcomes or singletons.summary(feglm_h3) ## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 14,656
## Fixed-effects: location_id: 916, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 2.411103 0.143424 16.810999 < 2.2e-16 ***
## log_funding_total 0.052493 0.005057 10.380346 < 2.2e-16 ***
## log_Total_Amount_Grants -0.003218 0.004152 -0.775203 0.43823300
## funding_count 0.000258 0.000316 0.817132 0.41386720
## Total_Grants 0.002408 0.000718 3.351928 0.00080468 ***
## unique_companies_count 0.000093 0.000245 0.378812 0.70483303
## d_ECO:log_funding_total -0.101260 0.007077 -14.308858 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.940719######## feglm_h4 (not significant) (d_f_feglm_hNS is significant)
feglm_h4 = feglm(log_Amount_I_Corps ~ d_HNS + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = diddata)## NOTE: 20,432 observations removed because of NA values (RHS: 20,432).summary(feglm_h4)## GLM estimation, family = gaussian, Dep. Var.: log_Amount_I_Corps
## Observations: 44,032
## Fixed-effects: location_id: 2,752, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_HNS -0.628372 0.095980 -6.546878 5.9440e-11 ***
## log_Total_Amount_Grants 0.002116 0.002235 0.947044 3.4362e-01
## log_funding_total 0.004989 0.001345 3.708125 2.0907e-04 ***
## funding_count 0.004957 0.002730 1.815916 6.9391e-02 .
## Total_Grants 0.015481 0.002324 6.661890 2.7372e-11 ***
## unique_companies_count 0.004915 0.002120 2.318708 2.0416e-02 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Log-Likelihood: -73,694.8 Adj. Pseudo R2: 0.119247
## BIC: 177,040.4 Squared Cor.: 0.449075#MAIN
feglm_h5 = feglm(companies_founded ~ log_Amount_I_Corps + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = diddata, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,836/0 fixed-effects (29,376 observations) removed because of only 0 outcomes or singletons.summary(feglm_h5)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 14,656
## Fixed-effects: location_id: 916, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## log_Amount_I_Corps 0.022860 0.002135 10.705135 < 2.2e-16 ***
## log_Total_Amount_Grants -0.003332 0.004180 -0.797064 4.2543e-01
## log_funding_total 0.037793 0.004702 8.036809 9.9701e-16 ***
## funding_count 0.000386 0.000317 1.216717 2.2373e-01
## Total_Grants 0.002084 0.000731 2.850982 4.3649e-03 **
## unique_companies_count -0.000196 0.000242 -0.809500 4.1824e-01
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.935604low_data_acc<-diddata %>% filter(Total_Amount_Grants_sum_acc ==0)######## low_acc_h1
low_acc_h1 = feglm(companies_founded ~ d_ECO + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = low_data_acc, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,835/0 fixed-effects (29,360 observations) removed because of only 0 outcomes or singletons.summary(low_acc_h1)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 13,792
## Fixed-effects: location_id: 862, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 0.606862 0.038646 15.70306 < 2.2e-16 ***
## log_Total_Amount_Grants -0.005703 0.004222 -1.35071 1.7681e-01
## log_funding_total 0.032305 0.005012 6.44571 1.1915e-10 ***
## funding_count 0.002565 0.000460 5.57457 2.5312e-08 ***
## Total_Grants 0.010593 0.002052 5.16205 2.4789e-07 ***
## unique_companies_count -0.002175 0.000370 -5.87961 4.2140e-09 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.871594######## low_acc_h2 & low_acc_h3 - moderations
#MAIN
low_acc_h2 = feglm(companies_founded ~ d_ECO * log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = low_data_acc, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,835/0 fixed-effects (29,360 observations) removed because of only 0 outcomes or singletons.summary(low_acc_h2)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 13,792
## Fixed-effects: location_id: 862, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 3.499095 0.429660 8.14388 4.1779e-16 ***
## log_Total_Amount_Grants -0.004297 0.004232 -1.01548 3.0990e-01
## log_funding_total 0.031955 0.005023 6.36199 2.0588e-10 ***
## funding_count 0.002026 0.000465 4.35854 1.3195e-05 ***
## Total_Grants 0.015069 0.002153 6.99935 2.6985e-12 ***
## unique_companies_count -0.001652 0.000376 -4.39983 1.0920e-05 ***
## d_ECO:log_Total_Amount_Grants -0.181462 0.026864 -6.75487 1.4909e-11 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.874559#MAIN
low_acc_h3 = feglm(companies_founded ~ d_ECO * log_funding_total + log_Total_Amount_Grants +
funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = low_data_acc, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,835/0 fixed-effects (29,360 observations) removed because of only 0 outcomes or singletons.summary(low_acc_h3) ## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 13,792
## Fixed-effects: location_id: 862, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 2.698015 0.167405 16.11673 < 2.2e-16 ***
## log_funding_total 0.048743 0.005410 9.00951 < 2.2e-16 ***
## log_Total_Amount_Grants -0.006088 0.004216 -1.44404 1.4875e-01
## funding_count 0.001249 0.000470 2.65634 7.9091e-03 **
## Total_Grants 0.012956 0.002062 6.28389 3.4083e-10 ***
## unique_companies_count -0.000760 0.000384 -1.97696 4.8068e-02 *
## d_ECO:log_funding_total -0.108653 0.008464 -12.83741 < 2.2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.886559######## low_acc_h4 (not significant) (d_f_hNS is significant)
low_acc_h4 = feols(log_Amount_I_Corps ~ d_HNS + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = low_data_acc)## NOTE: 20,432 observations removed because of NA values (RHS: 20,432).summary(low_acc_h4)## OLS estimation, Dep. Var.: log_Amount_I_Corps
## Observations: 43,152
## Fixed-effects: location_id: 2,697, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_HNS -0.298725 0.097511 -3.06350 2.1891e-03 **
## log_Total_Amount_Grants -0.007852 0.001808 -4.34411 1.4018e-05 ***
## log_funding_total 0.003692 0.001075 3.43339 5.9668e-04 ***
## funding_count -0.003903 0.002979 -1.31018 1.9014e-01
## Total_Grants 0.093365 0.003308 28.22452 < 2.2e-16 ***
## unique_companies_count 0.008122 0.002360 3.44195 5.7812e-04 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## RMSE: 1.05775 Adj. R2: 0.326394
## Within R2: 0.023903#MAIN
low_acc_h5 = feglm(companies_founded ~ log_Amount_I_Corps + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = low_data_acc, family = "quasipoisson")## NOTES: 20,432 observations removed because of NA values (RHS: 20,432).
## 1,835/0 fixed-effects (29,360 observations) removed because of only 0 outcomes or singletons.summary(low_acc_h5)## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 13,792
## Fixed-effects: location_id: 862, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## log_Amount_I_Corps 0.039959 0.003331 11.99497 < 2.2e-16 ***
## log_Total_Amount_Grants -0.005913 0.004234 -1.39658 1.6256e-01
## log_funding_total 0.032884 0.005012 6.56143 5.5329e-11 ***
## funding_count 0.001502 0.000458 3.27833 1.0470e-03 **
## Total_Grants 0.010512 0.002069 5.08085 3.8101e-07 ***
## unique_companies_count -0.001040 0.000363 -2.86919 4.1220e-03 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.866921medium_data_acc<-diddata %>% filter(Total_Amount_Grants_sum_acc ==1)####0
medium_acc_h1 <- feglm(log_companies_founded ~ d_ECO + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count , data = medium_data_acc, effect="twoways", index=c("location_id", "AwardEffectiveDate"),
model="within") #“within”, “random”, “medium_ht”, “between”, “pooling”,## Warning: In fixest_env(fml = fml, data = data, family = famil...:
## feglm(fml = log...: effect, index and model are not valid arguments for
## function feglm()#coeftest(medium_acc_h1, vcov=vcovmedium_hC, type="medium_hC3")
summary(medium_acc_h1)## GLM estimation, family = gaussian, Dep. Var.: log_companies_founded
## Observations: 608
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.782095 1.707399 2.215121 2.7126e-02 *
## d_ECO 0.127182 0.079207 1.605696 1.0887e-01
## log_Total_Amount_Grants -0.260906 0.100712 -2.590627 9.8126e-03 **
## log_funding_total 0.078818 0.007419 10.623820 < 2.2e-16 ***
## funding_count 0.011988 0.004736 2.531149 1.1623e-02 *
## Total_Grants 0.006563 0.001267 5.180619 3.0229e-07 ***
## unique_companies_count -0.003165 0.003980 -0.795083 4.2688e-01
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Log-Likelihood: -804.8 Adj. Pseudo R2: 0.215492
## BIC: 1,654.4 Squared Cor.: 0.528723######## medium_h2 & medium_h3 - moderations
#MAIN
medium_acc_h2 = feglm(log_companies_founded ~ d_ECO * log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count , data = medium_data_acc, effect="twoways", index=c("location_id", "AwardEffectiveDate"),
model="within") #“within”, “random”, “medium_acc_ht”, “between”, “pooling”,## Warning: In fixest_env(fml = fml, data = data, family = famil...:
## feglm(fml = log...: effect, index and model are not valid arguments for
## function feglm()summary(medium_acc_h2)## GLM estimation, family = gaussian, Dep. Var.: log_companies_founded
## Observations: 608
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.295596 2.218721 0.583938 5.5948e-01
## d_ECO 5.253536 2.926077 1.795419 7.3090e-02 .
## log_Total_Amount_Grants -0.120062 0.128773 -0.932349 3.5153e-01
## log_funding_total 0.078345 0.007417 10.562767 < 2.2e-16 ***
## funding_count 0.011211 0.004753 2.358936 1.8647e-02 *
## Total_Grants 0.006647 0.001267 5.247592 2.1408e-07 ***
## unique_companies_count -0.002344 0.004004 -0.585338 5.5854e-01
## d_ECO:log_Total_Amount_Grants -0.290049 0.165497 -1.752595 8.0182e-02 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Log-Likelihood: -803.2 Adj. Pseudo R2: 0.216044
## BIC: 1,657.7 Squared Cor.: 0.531152#MAIN
medium_acc_h3 = feglm(log_companies_founded ~ d_ECO * log_funding_total +
log_Total_Amount_Grants + funding_count + Total_Grants + unique_companies_count , data = medium_data_acc, effect="twoways", index=c("location_id", "AwardEffectiveDate"),
model="within") #“within”, “random”, “medium_acc_ht”, “between”, “pooling”,## Warning: In fixest_env(fml = fml, data = data, family = famil...:
## feglm(fml = log...: effect, index and model are not valid arguments for
## function feglm()summary(medium_acc_h3) ## GLM estimation, family = gaussian, Dep. Var.: log_companies_founded
## Observations: 608
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.857156 1.711432 2.253759 2.4571e-02 *
## d_ECO -0.097726 0.274277 -0.356305 7.2174e-01
## log_funding_total 0.075664 0.008289 9.127772 < 2.2e-16 ***
## log_Total_Amount_Grants -0.262505 0.100834 -2.603323 9.4609e-03 **
## funding_count 0.012245 0.004751 2.577598 1.0186e-02 *
## Total_Grants 0.006573 0.001268 5.182477 2.9957e-07 ***
## unique_companies_count -0.003531 0.004007 -0.881037 3.7865e-01
## d_ECO:log_funding_total 0.013788 0.016097 0.856576 3.9202e-01
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Log-Likelihood: -804.4 Adj. Pseudo R2: 0.214888
## BIC: 1,660.0 Squared Cor.: 0.529306######## medium_acc_h4 (not significant) (d_f_medium_acc_hNS is significant)
medium_acc_h4 = feglm(log_Amount_I_Corps ~ d_HNS + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count , data = medium_data_acc, effect="twoways", index=c("location_id", "AwardEffectiveDate"),
model="within") #“within”, “random”, “medium_acc_ht”, “between”, “pooling”,## Warning: In fixest_env(fml = fml, data = data, family = famil...:
## feglm(fml = log...: effect, index and model are not valid arguments for
## function feglm()summary(medium_acc_h4)## GLM estimation, family = gaussian, Dep. Var.: log_Amount_I_Corps
## Observations: 608
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.895649 10.150423 0.186756 8.5192e-01
## d_HNS -1.439736 0.736845 -1.953922 5.1174e-02 .
## log_Total_Amount_Grants -0.114179 0.598608 -0.190741 8.4879e-01
## log_funding_total 0.169766 0.043044 3.944025 8.9572e-05 ***
## funding_count -0.035763 0.028032 -1.275787 2.0252e-01
## Total_Grants 0.023951 0.007520 3.184940 1.5228e-03 **
## unique_companies_count 0.037537 0.023536 1.594843 1.1127e-01
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Log-Likelihood: -1,888.1 Adj. Pseudo R2: 0.014368
## BIC: 3,821.0 Squared Cor.: 0.10467#MAIN
medium_acc_h5 = feglm(log_companies_founded ~ log_Amount_I_Corps + log_Total_Amount_Grants +
log_funding_total + funding_count + Total_Grants + unique_companies_count , data = medium_data_acc, effect="twoways", index=c("location_id", "AwardEffectiveDate"),
model="within") #“within”, “random”, “medium_acc_ht”, “between”, “pooling”,## Warning: In fixest_env(fml = fml, data = data, family = famil...:
## feglm(fml = log...: effect, index and model are not valid arguments for
## function feglm()summary(medium_acc_h5)## GLM estimation, family = gaussian, Dep. Var.: log_companies_founded
## Observations: 608
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.636979 1.702745 2.135950 3.3087e-02 *
## log_Amount_I_Corps 0.015576 0.006864 2.269128 2.3614e-02 *
## log_Total_Amount_Grants -0.251582 0.100420 -2.505303 1.2498e-02 *
## log_funding_total 0.078922 0.007303 10.806455 < 2.2e-16 ***
## funding_count 0.011896 0.004715 2.523052 1.1891e-02 *
## Total_Grants 0.006274 0.001274 4.926214 1.0852e-06 ***
## unique_companies_count -0.003141 0.003962 -0.792883 4.2816e-01
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Log-Likelihood: -803.5 Adj. Pseudo R2: 0.216755
## BIC: 1,651.8 Squared Cor.: 0.530742high_data_acc<-diddata %>% filter(Total_Amount_Grants_sum_acc ==2)high_acc_h1 = feglm(companies_founded ~ d_ECO + Total_Amount_Grants +
funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = high_data_acc, family = "quasipoisson")## NOTE: 1/0 fixed-effect (16 observations) removed because of only 0 outcomes or singletons.summary(high_acc_h1)## Warning: The VCOV matrix is not positive semi-definite and was 'fixed' (see
## ?vcov).## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 256
## Fixed-effects: location_id: 16, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 7.004432e-02 0.06236678 1.12310297 0.26262
## Total_Amount_Grants -6.404400e-10 0.00000100 -0.00064044 0.99949
## funding_total 5.520000e-12 0.00000100 0.00000552 1.00000
## funding_count -6.162102e-04 0.00055819 -1.10393539 0.27083
## Total_Grants 9.415256e-05 0.00111042 0.08479016 0.93251
## unique_companies_count 3.548796e-04 0.00045152 0.78596923 0.43274
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.9799######## high_acc_h2 & high_acc_h3 - moderations
#MAIN
high_acc_h2 = feglm(companies_founded ~ d_ECO * Total_Amount_Grants +
funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = high_data_acc, family = "quasipoisson")## NOTE: 1/0 fixed-effect (16 observations) removed because of only 0 outcomes or singletons.summary(high_acc_h2)## Warning: The VCOV matrix is not positive semi-definite and was 'fixed' (see
## ?vcov).## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 256
## Fixed-effects: location_id: 16, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 9.307736e-04 0.09533584 0.00976310 0.99222
## Total_Amount_Grants -1.077490e-09 0.00000100 -0.00107749 0.99914
## funding_total 5.700000e-12 0.00000100 0.00000570 1.00000
## funding_count -5.854982e-04 0.00056296 -1.04004257 0.29947
## Total_Grants 6.671826e-05 0.00111822 0.05966488 0.95248
## unique_companies_count 3.208945e-04 0.00045591 0.70385521 0.48227
## d_ECO:Total_Amount_Grants 6.106400e-10 0.00000100 0.00061064 0.99951
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.979891#MAIN
high_acc_h3 = feglm(companies_founded ~ d_ECO * funding_total + Total_Amount_Grants +
funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = high_data_acc, family = "quasipoisson")## NOTE: 1/0 fixed-effect (16 observations) removed because of only 0 outcomes or singletons.summary(high_acc_h3) ## Warning: The VCOV matrix is not positive semi-definite and was 'fixed' (see
## ?vcov).## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 256
## Fixed-effects: location_id: 16, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_ECO 2.501396e-01 0.08727962 2.865956 0.0045646 **
## funding_total 1.290000e-10 0.00000100 0.000129 0.9998973
## Total_Amount_Grants -4.510000e-10 0.00000100 -0.000451 0.9996403
## funding_count -6.360494e-04 0.00055447 -1.147128 0.2525860
## Total_Grants -1.104573e-03 0.00117312 -0.941569 0.3474557
## unique_companies_count 6.345931e-04 0.00045835 1.384502 0.1676202
## d_ECO:funding_total -1.260000e-10 0.00000100 -0.000126 0.9998998
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.980726######## high_acc_h4 (not significant) (d_f_high_acc_hNS is significant)
high_acc_h4 = feglm(Amount_I_Corps ~ d_HNS + Total_Amount_Grants +
funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = high_data_acc, family = "quasipoisson")## NOTE: 1/3 fixed-effects (64 observations) removed because of only 0 outcomes or singletons.summary(high_acc_h4)## Warning: The VCOV matrix is not positive semi-definite and was 'fixed' (see
## ?vcov).## GLM estimation, family = quasipoisson, Dep. Var.: Amount_I_Corps
## Observations: 208
## Fixed-effects: location_id: 16, AwardEffectiveDate: 13
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## d_HNS -5.758951e-01 0.22338441 -2.57804532 0.0107640 *
## Total_Amount_Grants -1.189000e-09 0.00000100 -0.00118900 0.9990527
## funding_total 4.360000e-12 0.00000100 0.00000436 0.9999965
## funding_count 2.973420e-03 0.00338288 0.87896111 0.3806346
## Total_Grants 1.289346e-02 0.00493335 2.61353225 0.0097461 **
## unique_companies_count -2.297374e-03 0.00271255 -0.84694174 0.3981910
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.42147#MAIN
high_acc_h5 = feglm(companies_founded ~ Amount_I_Corps + Total_Amount_Grants +
funding_total + funding_count + Total_Grants + unique_companies_count | location_id + AwardEffectiveDate, data = high_data_acc, family = "quasipoisson")## NOTE: 1/0 fixed-effect (16 observations) removed because of only 0 outcomes or singletons.summary(high_acc_h5)## Warning: The VCOV matrix is not positive semi-definite and was 'fixed' (see
## ?vcov).## GLM estimation, family = quasipoisson, Dep. Var.: companies_founded
## Observations: 256
## Fixed-effects: location_id: 16, AwardEffectiveDate: 16
## Standard-errors: IID
## Estimate Std. Error t value Pr(>|t|)
## Amount_I_Corps -8.351938e-08 0.00000100 -0.08347540 0.93355
## Total_Amount_Grants -6.259700e-10 0.00000100 -0.00062597 0.99950
## funding_total 5.280000e-12 0.00000100 0.00000528 1.00000
## funding_count -5.666949e-04 0.00056444 -1.00398757 0.31649
## Total_Grants 2.545217e-04 0.00114207 0.22285925 0.82385
## unique_companies_count 3.370467e-04 0.00045475 0.74117416 0.45938
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Squared Cor.: 0.979585