Why multiple statements executed in single run takes more time than when those statement run individually?

Question

I am running following queries from DBeaver app connected to Postgres.

Statement1:

-- create 1000 rows in table2

do $$
    declare 
        begin
            for r in 1..1000 loop
                insert into table2 (id, table1_id, table2_col3, table2_col4, table2_col5) 
                values ( uuid_generate_v4(), '5d353232-aa7e-4556-9cd3-3bdae43c25d4' , 'data', 0, true);
            end loop;
        end; 
    $$;

Statement 2 and 3 create data in table3 based upon the rows in table2 (table3 have FK-PK relation to table2)

Statement 2:

insert into table3 (
    id, table2_id , table3_col3 , table3_col4
    )
    select uuid_generate_v4(), id , 'data' ,  'another data' from table2 where table1_id = '5d353232-aa7e-4556-9cd3-3bdae43c25d4'
    on conflict do nothing ;

Statement 3:

insert into table3 (
id, table2_id , table3_col3 , table3_col4
)
select uuid_generate_v4(), id , 'different data' ,  'another different data' from table2 where table1_id = '5d353232-aa7e-4556-9cd3-3bdae43c25d4'
on conflict do nothing ;

So after the call finishes, I will have 1000 rows in table2 and 2000 rows in table3.

The question I have is, when I execute Statement 2 and 3 individually, it takes far less time(1/3) than when I run by selecting both statements and click execute button. What could be the reason behind it?

I have experienced similar problem when I have written the above statements inside Stored Procedure and made a call to it.

PS: The code written is for creating sample data in database.

Edit 1: Table DDL statements

CREATE TABLE IF NOT EXISTS table1
(
  id                    UUID        PRIMARY KEY,
);

CREATE TABLE IF NOT EXISTS table2
(
  id                    UUID        PRIMARY KEY,
  table1_id         UUID        NOT NULL REFERENCES table1,
  table2_col3                   TEXT        NOT NULL,
  table2_col4           INT,
  table2_col5               BOOLEAN     NOT NULL DEFAULT TRUE
);

CREATE TABLE IF NOT EXISTS table3
(
  id                    UUID        PRIMARY KEY,
  table2_id     UUID        REFERENCES  table2,
  table3_col3           VARCHAR(7)  NOT NULL,
  table3_col4           TEXT        NOT NULL,
);

Edit 2:

I had run EXPLAIN (ANALYZE, BUFFERS) STATEMENT1; STATEMENT2; Following is the output:

Insert on table3  (cost=0.00..47810.00 rows=0 width=0) (actual time=1431686.852..1431686.853 rows=0 loops=1)
  Conflict Resolution: NOTHING
  Tuples Inserted: 1000000
  Conflicting Tuples: 0
  Buffers: shared hit=19518252 read=4863344 dirtied=3647588 written=2997004
  ->  Subquery Scan on "*SELECT*"  (cost=0.00..47810.00 rows=1000000 width=96) (actual time=0.701..20941.791 rows=1000000 loops=1)
        Buffers: shared hit=2 read=10308 written=3064
        ->  Seq Scan on table2  (cost=0.00..37810.00 rows=1000000 width=96) (actual time=0.700..20458.909 rows=1000000 loops=1)
              Filter: (table1_id = '7a528463-1133-4720-8550-1760703e1717'::uuid)
              Buffers: shared hit=2 read=10308 written=3064
Planning:
  Buffers: shared hit=62 read=3 dirtied=1
Planning Time: 3.310 ms
Trigger for constraint table3_table2_id_fkey: time=19972.740 calls=1000000
Execution Time: 1451736.223 ms

After that I ran EXPLAIN (ANALYZE, BUFFERS) STATEMENT1; and I get following output:

Insert on table3  (cost=0.00..47810.00 rows=0 width=0) (actual time=1502673.065..1502673.065 rows=0 loops=1)
  Conflict Resolution: NOTHING
  Tuples Inserted: 1000000
  Conflicting Tuples: 0
  Buffers: shared hit=19404094 read=4964519 dirtied=3746289 written=3042279
  ->  Subquery Scan on "*SELECT*"  (cost=0.00..47810.00 rows=1000000 width=96) (actual time=0.318..19945.951 rows=1000000 loops=1)
        Buffers: shared read=10310 written=3462
        ->  Seq Scan on table2  (cost=0.00..37810.00 rows=1000000 width=96) (actual time=0.317..19435.714 rows=1000000 loops=1)
              Filter: (table1_id = '7a528463-1133-4720-8550-1760703e1717'::uuid)
              Buffers: shared read=10310 written=3462
Planning Time: 0.066 ms
Trigger for constraint table3_table2_id_fkey: time=18026.509 calls=1000000
Execution Time: 1520767.185 ms

I feel that in first case, it shows output of first or second statement not both, because the times are almost similar.