--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
      name:  <unnamed>
       log:  D:/stata17/Replication_paper/Logfile_forReg/Analysis_产业扩张的区位差异_新进入企业.txt
  log type:  text
 opened on:  25 Jan 2026, 12:06:17

. 
. ***************
. * Run
. ***************
. 
. 
. use "$DATA\intermediate\newfirms_county_reviseRegcap.dta",clear

. 
. * reshape newfirms_manuf (0/1 -> below/high)
. rename new_firms_manuf_above new_firms_manuf1

. rename new_firms_manuf_below new_firms_manuf0

. rename new_firms_modsrv new_firms_service1

. rename new_firms_basesrv new_firms_service0

. 
. rename regcap_manuf_above regcap_manuf1

. rename regcap_manuf_below regcap_manuf0

. rename regcap_modsrv regcap_service1

. rename regcap_basesrv regcap_service0

. reshape long new_firms_manuf new_firms_service regcap_manuf regcap_service,i(year domdistrict) j(group)
(j = 0 1)

Data                               Wide   ->   Long
-----------------------------------------------------------------------------
Number of observations           29,874   ->   59,748      
Number of variables                  10   ->   7           
j variable (2 values)                     ->   group
xij variables:
      new_firms_manuf0 new_firms_manuf1   ->   new_firms_manuf
  new_firms_service0 new_firms_service1   ->   new_firms_service
            regcap_manuf0 regcap_manuf1   ->   regcap_manuf
        regcap_service0 regcap_service1   ->   regcap_service
-----------------------------------------------------------------------------

. label define grouplab 0 "制造业低关联度/非试点服务业" 1 "制造业高关联度/试点现代服务业"

. label values group grouplab

. 
. * 1 combine variable
. * combine 县区人口密度
. gen long domdistrict_num = real(domdistrict)
(102 missing values generated)

. rename domdistrict_num countycode

. merge m:1 countycode using "$DATA\sourcedata\2010popdensity.dta "

    Result                      Number of obs
    -----------------------------------------
    Not matched                        27,006
        from master                    26,984  (_merge==1)
        from using                         22  (_merge==2)

    Matched                            32,764  (_merge==3)
    -----------------------------------------

. keep if _merge == 3
(27,006 observations deleted)

. drop _merge

. 
. * combine control variable
. *县区人均gdp, 产业结构(三产/二产)
. rename year years

. rename countycode xx

. merge m:1 xx years using "$DATA\sourcedata\区县控制变量_带Label.dta"

    Result                      Number of obs
    -----------------------------------------
    Not matched                         9,600
        from master                       696  (_merge==1)
        from using                      8,904  (_merge==2)

    Matched                            32,068  (_merge==3)
    -----------------------------------------

. keep if _merge == 3
(9,600 observations deleted)

. drop _merge

. 
. gen gdpper = gdp/pophuji   // gdp/户籍人口
(2,778 missing values generated)

. gen industryratio = thirdgdp/secondgdp  //产业结构 = 三产/二产
(2,248 missing values generated)

. label var pophuji "户籍人口,表示人口规模"

. 
. * combine 全国城市群设定,筛选全国城市群样本
. rename xx countycode

. merge m:1 countycode using "$DATA\sourcedata\distanceBetweenCounty.dta "

    Result                      Number of obs
    -----------------------------------------
    Not matched                         8,339
        from master                     8,266  (_merge==1)
        from using                         73  (_merge==2)

    Matched                            23,802  (_merge==3)
    -----------------------------------------

. keep if _merge == 3
(8,339 observations deleted)

. drop _merge

. replace km = km - 1 //调整县区到所在中心县区的距离基准点为0，原有数据+ 1km，现在调整回去
(23,802 real changes made)

. 
. * 2  设定"营改增"政策变量
. gen prov = real(substr(domdistrict,1,2))

. gen int policy_year = .
(23,802 missing values generated)

. replace policy_year = 2012 if prov == 31
(168 real changes made)

. replace policy_year = 2013 if inlist(prov, 11,32, 33, 34, 42,12,35,44)
(6,298 real changes made)

. replace policy_year = 2014 if inlist(prov, 13,14,15,21,22,23,36,37,41,43,45,46,50,51,52,53,54,61,62,63,64,65)
(17,336 real changes made)

. 
. * 生成来源地政策 dummy：实施年份及以后为 1，之前为 0
. gen byte policy = (year >= policy_year) 

. count if missing(policy)
  0

. * 检查
. tab prov policy_year

           |           policy_year
      prov |      2012       2013       2014 |     Total
-----------+---------------------------------+----------
        11 |         0        192          0 |       192 
        12 |         0        180          0 |       180 
        13 |         0          0      1,934 |     1,934 
        14 |         0          0        996 |       996 
        15 |         0          0        312 |       312 
        21 |         0          0        780 |       780 
        22 |         0          0        498 |       498 
        23 |         0          0        516 |       516 
        31 |       168          0          0 |       168 
        32 |         0      1,016          0 |     1,016 
        33 |         0      1,002          0 |     1,002 
        34 |         0      1,176          0 |     1,176 
        35 |         0        976          0 |       976 
        36 |         0          0      1,058 |     1,058 
        37 |         0          0        720 |       720 
        41 |         0          0      1,202 |     1,202 
        42 |         0      1,236          0 |     1,236 
        43 |         0          0      1,464 |     1,464 
        44 |         0        520          0 |       520 
        45 |         0          0        434 |       434 
        50 |         0          0        432 |       432 
        51 |         0          0      2,162 |     2,162 
        52 |         0          0      1,028 |     1,028 
        53 |         0          0      1,540 |     1,540 
        61 |         0          0        870 |       870 
        62 |         0          0        504 |       504 
        63 |         0          0        142 |       142 
        64 |         0          0        204 |       204 
        65 |         0          0        540 |       540 
-----------+---------------------------------+----------
     Total |       168      6,298     17,336 |    23,802 

. tab policy

     policy |      Freq.     Percent        Cum.
------------+-----------------------------------
          0 |     18,708       78.60       78.60
          1 |      5,094       21.40      100.00
------------+-----------------------------------
      Total |     23,802      100.00

. * 若 policy_year 缺失，会让 policy 变成缺失；这里把缺失置为0更稳妥
. replace policy = 0 if missing(policy)
(0 real changes made)

. 
. * 3 regress 
. ***************regression: 表3(3).(4)列:产业扩张的区位差异—新企业进入*****************
. 
. rename years year

. rename bb citycluster

. label var citycluster "区分全国城市群单元"

. rename km distance

. replace distance = distance/1000  //变量缩放1000倍,以便统一小数点后3位数
(23,550 real changes made)

. 
. gen policy_group = policy * group 

. gen lnnewfirm_service = ln(1 + new_firms_service)

. gen lnnewfirm_manuf = ln(1 + new_firms_manuf)

. 
. local controlVar "gdpper  pophuji"

. reghdfe lnnewfirm_manuf  i.policy_group##c.distance `controlVar' , absorb(countycode year  )  cluster(countycode) 
(MWFE estimator converged in 5 iterations)
note: distance is probably collinear with the fixed effects (all partialled-out values are close to zero; tol = 1.0e-09)

HDFE Linear regression                            Number of obs   =     22,196
Absorbing 2 HDFE groups                           F(   4,   1942) =      40.02
Statistics robust to heteroskedasticity           Prob > F        =     0.0000
                                                  R-squared       =     0.8396
                                                  Adj R-squared   =     0.8242
                                                  Within R-sq.    =     0.0175
Number of clusters (countycode) =      1,943      Root MSE        =     0.5730

                                    (Std. err. adjusted for 1,943 clusters in countycode)
-----------------------------------------------------------------------------------------
                        |               Robust
        lnnewfirm_manuf | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
------------------------+----------------------------------------------------------------
         1.policy_group |  -.2534469   .0242561   -10.45   0.000    -.3010177    -.205876
               distance |          0  (omitted)
                        |
policy_group#c.distance |
                     1  |   .7483057    .077828     9.61   0.000     .5956705     .900941
                        |
                 gdpper |  -4.92e-06   1.04e-06    -4.72   0.000    -6.97e-06   -2.88e-06
                pophuji |   .0008499   .0007142     1.19   0.234    -.0005507    .0022505
                  _cons |   2.873044   .0614624    46.74   0.000     2.752505    2.993583
-----------------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
  countycode |      1943        1943           0    *|
        year |         6           1           5     |
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

. estimates store m1

. reghdfe lnnewfirm_service i.policy_group##c.distance `controlVar' , absorb(  countycode year  ) cluster(countycode)
(MWFE estimator converged in 5 iterations)
note: distance is probably collinear with the fixed effects (all partialled-out values are close to zero; tol = 1.0e-09)

HDFE Linear regression                            Number of obs   =     22,196
Absorbing 2 HDFE groups                           F(   4,   1942) =      16.56
Statistics robust to heteroskedasticity           Prob > F        =     0.0000
                                                  R-squared       =     0.9034
                                                  Adj R-squared   =     0.8941
                                                  Within R-sq.    =     0.0054
Number of clusters (countycode) =      1,943      Root MSE        =     0.4736

                                    (Std. err. adjusted for 1,943 clusters in countycode)
-----------------------------------------------------------------------------------------
                        |               Robust
      lnnewfirm_service | Coefficient  std. err.      t    P>|t|     [95% conf. interval]
------------------------+----------------------------------------------------------------
         1.policy_group |   .1240903   .0204296     6.07   0.000      .084024    .1641566
               distance |          0  (omitted)
                        |
policy_group#c.distance |
                     1  |  -.3693608   .0554139    -6.67   0.000    -.4780377   -.2606839
                        |
                 gdpper |  -2.47e-06   6.22e-07    -3.96   0.000    -3.69e-06   -1.25e-06
                pophuji |   .0010019   .0004924     2.03   0.042     .0000363    .0019676
                  _cons |   3.748382   .0382743    97.93   0.000     3.673319    3.823445
-----------------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
  countycode |      1943        1943           0    *|
        year |         6           1           5     |
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

. estimates store m2

. 
. 
. *  m1 m2导出，替换已有文件
. outreg2 [m1] using "$EXHIBIT/results.doc", replace ///
>     ctitle("newfirms_manuf")          /// 列标题
>     se                       /// 报告括号内标准误
>     dec(3)  nocons addtext(County FE, Yes,Year FE, Yes)                  
D:/stata17/Replication_paper/Exhibits/results.doc
dir : seeout

. 
. outreg2 [m2] using "$EXHIBIT/results.doc", append ///
>     ctitle("newfirms_service")          /// 列标题
>     se                       /// 报告括号内标准误
>     dec(3)   nocons addtext(County FE, Yes,Year FE,Yes)                  
D:/stata17/Replication_paper/Exhibits/results.doc
dir : seeout

. 
. 
. 
. 
. *****************
. * close log txt
. *****************
. 
. log close
      name:  <unnamed>
       log:  D:/stata17/Replication_paper/Logfile_forReg/Analysis_产业扩张的区位差异_新进入企业.txt
  log type:  text
 closed on:  25 Jan 2026, 12:06:18
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
