How to deal with error correlation model with Stata

[Sc Meng]

regression function is :

In the following data, variable value is the value of 1952, pi1952 is the price index based on 1952
and the time range is from 1953-1978. So, I drop data if year is 1952 and 1990 after changing based price using 1990 price index.
note: when t=0, it means 1978 and t-1 is 1977.

data:
[CODE]
* Example generated by -dataex-. To install: ssc install dataex
clear
input int t double(value pi1952)
1952 80.7 1
1953 115.3 .988
1954 140.9 .982
1955 145.5 .94
1956 219.6 .937
1957 187 .897
1958 333 .901
1959 435.7 .976
1960 473 .973
1961 227.6 .955
1962 175.1 1.025
1963 215.3 1.075
1964 290.3 1.053
1965 350.1 1.018
1966 406.8 .998
1967 323.7 1.002
1968 300.2 .967
1969 406.9 .945
1970 545.9 .945
1971 603 .955
1972 622.1 .967
1973 664.5 .968
1974 748.1 .969
1975 880.3 .981
1976 865.1 .988
1977 911.1 1.002
1978 1073.9 1.008
1990 . 1.862
end

I change the value to 1990 price, and related codes are as following:
gen index=1/1.862
gen pi1990=pi1952*index
gen year=t-1978
drop in 1
tsset year,yearly
replace value=value/pi1990
gen lv=log(value)
drop if t==1990
reg lv year
predict e,res
twoway (scatter e e2)(lfit e e2)
twoway (scatter e e3)(lfit e e3)
estat bgo // Prob > chi2 0.0036<0.05
wntestq e1 //Prob > chi2(11) = 0.0469
di 26^(1/4) // and gain 3

newey lv year,lag(3)

I actually regress the data and cannot obtain the same coefficients as the author.
I don't know what problems I have in my codes, please help me.

and my results:

but the author gains coefficients like this:

[Sc Meng]

* Example generated by -dataex-. To install: ssc install dataex
clear
input int t double(value pi1952)
1952   80.7     1
1953  115.3  .988
1954  140.9  .982
1955  145.5   .94
1956  219.6  .937
1957    187  .897
1958    333  .901
1959  435.7  .976
1960    473  .973
1961  227.6  .955
1962  175.1 1.025
1963  215.3 1.075
1964  290.3 1.053
1965  350.1 1.018
1966  406.8  .998
1967  323.7 1.002
1968  300.2  .967
1969  406.9  .945
1970  545.9  .945
1971    603  .955
1972  622.1  .967
1973  664.5  .968
1974  748.1  .969
1975  880.3  .981
1976  865.1  .988
1977  911.1 1.002
1978 1073.9 1.008
1990      . 1.862
end

[Sc Meng]

Waiting for help

[Sc Meng]

This regression seems to be different from common AR models because it just needs to regress the value(y) with time trend(x) but not other independent variable. Meanwhile, one should address the autocorrelation problem. Thanks very for you help in advance

[Sc Meng]

gen index=1/1.862 //
gen pi1990=pi1952*index
gen year=_n
drop in 1
order year
tsset year,yearly
replace value=value/pi1990
gen lv=log(value)
//gen lv1=log(value1)
drop if t==1990
dfuller lv, trend //t= -2.339> -3.600
di 26^(1/4) //2.2581009
dfuller lv,lags(3) reg trend //
dfuller lv,lags(2) reg trend //
dfuller lv,lags(1) reg trend // statistically significant
pperron lv //
dfgls lv //
kpss lv, trend //

how can I continue my work, thanks

[Sc Meng]

um.....