if it's really your first time coding, then that's great: you actually managed to use some parts of the Python standard library that people being exposed for the first time to the language tend to needlessly reinvent, such as the csv module :)

Python version

The first big question: you used Python, but why did you use the version 2.7? This version (minus the bugfixes) is already 5 years old and its maintenance should have been dropped this year (it has been extended to 2020 since it's still widely used). But if you start from scratch, you might as well start with a fresh new version, such as Python 3.4. Generally speaking, don't use Python 2.7 unless you're forced to (there are still reasons); try to adapt to the most recent version instead.

Commented out code

Generally speaking, you should never leave commented-out pieces of code in your code. They will only hinder readability. I must admit that you probably don't use source control source control software if it's your first time programming, but you might want to look at software like Git in the future so that every version of your code is saved somewhere and you don't have to keep outdated remnants of code in your source.

Useless stuff in general

More generally, remove from your code what is unneeded, such as unused variables or unused includes. In your case, you could get rid of the re, os and time imports: you don't use re, you only use os in commented-out code that shouldn't belong, and what you use from time is actually also available in datetime.

The with statement

One thing that is easy to forget is closing files. You didn't forget to close them, which is great, but Python provides a superior alternative to closing files in the form of the with statement. Here are your first lines of code rewritten with it:

with open("Welcome-"+yesterday.strftime('%Y%m%d')+".csv","wb") as f:
    output=csv.writer(f)

    for row in new_data:
        output.writerow(row)

As you can see, I didn't manually call f.close(), this is automatically done when we leave the with block. Actually, with isn't limited to closing files, it can do many other things, but it depends on the type you use it with.

Use the PEP 8

Python has a style guide which is rather good and people tend to follow it. Therefore, you could try to rewrite your code following that style guide so that it looks more like what people want to read when they read Python code. For example, I will rewrite once again the previous piece of code, following the PEP8:

with open("Welcome-" + yesterday.strftime('%Y%m%d') + ".csv", "wb") as f:
    output = csv.writer(f)

    for row in new_data:
        output.writerow(row)

Basically, it's still the same. I only added spaces. But even those can matter when you read code :)

Consider using a database

The best kind of code is code that you don't have to write.

A completely different approach to this problem is not to solve it in Python, but to take advantage of SQL. Your three CSV files are basically tables in a relational database. Joining tables is a very routine thing in SQL.

If you have no database at all, the quickest way to get started would be with SQLite. A setup to solve your problem would look something like this (I've used placeholders for irrelevant column names):

CREATE TABLE SendJobs (Zero, ID, Two, Three, Four, Five, Six, Subject);

CREATE TABLE Attributes (Zero, One, EmailAddress, Three, Merchant_Number);

CREATE TABLE Bounces
( Zero, SendID, SubscriberKey, EmailAddress, Four, Five, Timestamp, Seven, Category, Nine, Reason
);

CREATE VIEW WelcomeBounces AS
    SELECT Bounces.SubscriberKey
         , Bounces.EmailAddress
         , Attributes.Merchant_Number
         , date(Bounces.Timestamp) AS EventDate
         , Bounces.Category AS BounceCategory
         , Bounces.Reason AS BounceReason
        FROM
            Bounces
                INNER JOIN SendJobs
                    ON SendJobs.ID = Bounces.SendID 
                INNER JOIN Attributes
                    ON Attributes.EmailAddress = Bounces.EmailAddress 
        WHERE
            SendJobs.Subject LIKE '%Welcome%';

Then, after importing the three CSV files, it's just a matter of executing

sqlite> .mode csv
sqlite> .header on
sqlite> .output Welcome.csv
sqlite> SELECT * FROM WelcomeBounces;

in the SQLite shell. Your problem is such a good match for SQL that it would be worthwhile to create and discard a temporary SQLite database just to solve it.