The effect of student loan size 
on outcomes of NZ bachelor’s 
graduates
Ran Abramitzky 
Stanford University and NBER 
Isabelle Sin  
Motu Economic and Public Policy Research and Te Pūnaha 
Matatini Centre of Research Excellence 
5 July 2019
This research was funded by a Marsden Grant 

 Motivation
Effect of student loan debt 
• $15.9b of NZ student debt at June 2018: 719,187 people 
with student loans, average owed $22,065 
• How are students’ futures and ability to utilise their 
education affected by debt? 
• Fees-Free education policy for first-time learners in 
force from 1 January 2018 
• How will resultant lower level of debt change behaviour 
of graduates?

 Research questions
How does student debt at degree completion affect the 
choices and outcomes of NZ bachelor’s graduates? 
1. Destination choice: overseas, further study, NZ job, 
other 
• Do loans drive grads overseas to earn higher salaries 
that will enable faster repayment, or to avoid 
repayment entirely? 
• Do loans encourage grads to stay in NZ to avoid 
interest? 
• Do loans discourage investment in higher 
qualifications?

 Research questions
2. Characteristics of first job:  
• whether graduate migrates to job 
• how soon after completion job starts  
• location relative to tertiary study and school location  
• sector (government, non-profit) 
• length of job spell  
• initial and subsequent earnings 

3. Family formation decisions:  
• marriage  
• parenthood

 Background: NZ’s student support system
• Financial support for tertiary students consists of loans 
(must be repaid) and allowances (not repaid) 
• Loans available for compulsory course fees (all 
students), living costs and course-related costs (FT only) 
• Repayment is income-contingent, 12c per dollar 
automatically deducted from income earned over a low 
threshold ($19,448 in April 2018) 
• Since 2006, no interest charged on loans for NZ-based 
borrowers, interest still charged for overseas-based 
borrowers

 Background: NZ’s student support system
• Student allowance eligibility based on parental income 
test for students under an age threshold 
- allowances help FT students cover living costs 
- threshold decreased from 25 years to 24 in 2009 
• Value of allowance abates $-for-$ once personal income 
increases over a threshold  
- so actual allowance received depends on choices 
such as how much to work while studying 
• Each dollar of allowance received reduces maximum 
borrowing for living expenses by $1  
- so students with increased allowance eligibility semimechanically borrow less

 SNZ Disclaimer
Access to the data presented was managed by 
Statistics New Zealand under strict micro-data access 
protocols and in accordance with the security and 
confidentiality provisions of the Statistics Act 1975.  
Our findings are not Official Statistics. The opinions, 
findings, recommendations, and conclusions 
expressed are those of the authors, not Statistics NZ, 
Motu Economic and Public Policy Research, Stanford 
University, or Te Punaha Matatini. 

 Data
• Integrated Data Infrastructure (IDI) held by Statistics 
New Zealand 
• Unit record data for whole of NZ based on administrative 
and survey data, linked for individuals between different 
sources  
• Main sources: 
- Tertiary education data from Ministry of Education 
(qualifications completed, characteristics of courses 
and study undertaken) 
- Employment and earnings data (monthly wage 
earnings) from Inland Revenue 
 -Border crossings data from passport swipes 
- Marriages and births data from DIA

 Data
• Main sample: Full-time, domestic students who 
completed a bachelor’s degree in NZ aged 20-30 
between Nov 2005 and June 2013 
• “First destination”, within 36 months of completion: 
- Overseas: overseas for 300 or more days 
- Study: enrolled full time in tertiary study for two or 
more consecutive half-years 
- NZ job: took a job in NZ with a starting monthly 
salary of $1000+, which lasted at least 3 months, and 
in which they hadn’t worked in their last 6 months 
studying

 Empirical strategy
• Loan size at degree completion endogenous - affected by 
parental support, financial attitude, work while studying, 
career intentions etc 
• Allowance receipt affects loan size, but also endogenous 
• Step further back and instrument for loan size with 
allowance eligibility 
• Instrument: fraction of 36 months leading up to 
completion in which allowance not parental income tested 
• Controls include fixed effects for year/month of 
completion and flexible age control 
• Identifying variation from 2009 change in age parental 
income testing ceases decreasing from 25 to 24

 Descriptive statistics
Outcome
Destination: Overseas
Destination: Study
Destination: NZ job
Migrated to job
Govt job
Non-profit job
Job start month
Job duration over 12 mon
Year 1 wages
Marriage within 6 years
Parenthood within 6 yrs

Mean
0.088
0.221
0.623
0.322
0.293
0.058
7.467
0.531
25,594
0.204
0.115

Std Dev.

8.471
28,856

Observations
85,425
85,425
85,425
64,722
68,142
68,142
68,205
68,205
85,425
69,513
60,018

  I

Motu ECONOMIC AND PUBLIC POLICY RESEARCH

Instrument affects allowance received: Age at
completion vs allowance received, pre and post

0 
I




15

10

  

   

5
I





Allowance received in last three calendar years
of study (real 2006 $0005Exact age at bachelor's completion

 

 

 

completed in 2005 -08 completed in 201 1'14

 

 

Effect of instrument on loan size
Dependent variable: Loan size at
completion (real 2006 $000s)

(1)
(2)
Parental income test exempt
-13.396*** -12.614***
(1.287)
(1.203)
Female
-1.348***
(0.123)
NZ born
-0.526**
(0.219)
Pass rate (0-1)
-29.909***
(0.828)
Year/month of study completion FE
Yes
Yes
Age at completion spline
Yes
Yes
Additional personal characteristics
No
Yes
Outcomes during study
No
No
RC of study & of school FE
No
No

(3)
-13.052***
(1.075)
-1.115***
(0.118)
-1.820***
(0.207)
-15.601***
(4.505)
Yes
Yes
Yes
Yes
No

(4)
-11.777***
(1.184)
-1.279***
(0.121)
-0.798***
(0.219)
-26.217***
(0.825)
Yes
Yes
Yes
No
Yes

R-Squared
0.145
0.211
0.298
0.242
Observations
85,404
85,335
85,335
85,335
Standard errors are robust. * p<0.10, ** p<0.05, *** p<0.01.
Parental income test exempt: Fraction of last 36 months studying exempt from student
allowance parental income test.

 IV: Does loan size affect first destination?
Dependent variable: destination given
in column header

Year/month of study completion FE
Age at completion spline
Additional controls

Overseas
Study
NZ job
-0.006
-0.006
-0.015
-0.007 0.036*
0.025
(0.014) (0.014) (0.016) (0.016) (0.021) (0.021)
-0.012***
-0.008**
0.031***
(0.003)
(0.004)
(0.004)
-0.028***
-0.014**
0.032***
(0.005)
(0.006)
(0.007)
0.036
0.252***
-0.379***
(0.076)
(0.091)
(0.126)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
No
Yes
No
Yes

R-Squared
Percentage of successes
Observations

0.009
8.8
85,404

0.071
8.8
85,335

0.035
22.1
85,404

0.158
22.1
85,335

0.016
62.3
85,404

0.112
62.3
85,335

95% CI on coefficient on Loan value

-0.033
0.021

-0.034
0.021

-0.047
0.016

-0.038
0.024

-0.006
0.078

-0.017
0.067

Loan value at completion ($0,000s)
Female
NZ born
Pass rate (0-1)

Standard errors are robust. * p<0.10, ** p<0.05, *** p<0.01. Additional controls: ethnicity FE, school decile FE, field
of study FE, time spent overseas during study, quadratic in years of tertiary study, quadratics in total EFTS passed and
and total EFTS failed for degree, months worked and monthly earnings in 3 years before degree completion, whether
married and whether a parent at completion and female interactions, studied at multiple institutions.

 IV: Does loan size affect job or earnings?
Dependent variable:
Loan value at completion
($0,000s)
Extra controls
Percentage of successes
Observations

Migrated to job

Govt job

Non-profit job

0.068*** 0.037 0.123*** 0.108*** -0.004
(0.025) (0.025) (0.028) (0.024) (0.014)
Yes
Yes
33.5
33.6
27.0
27.0
5.6
50,403 50,382 53,145 53,127 53,145

0.006
(0.014)
Yes
5.6
53,127

Job duration over
Year 1 wages
12 mon
($000s)
Loan value at completion
0.017
0.187 0.063** 0.031 8.760*** 9.299***
($0,000s)
(0.379) (0.356) (0.027) (0.026) (1.124) (0.969)
Extra controls
Yes
Yes
Yes
Percentage of successes
54.9
54.9
Observations
53,184 53,169 53,184 53,169 125,310 125,238
Each coefficient from a separate IV regression. Standard errors are robust. * p<0.10, **
p<0.05, *** p<0.01. Minimal and extra controls as previously. Sample for first five dep
vars is students whose first destination was a NZ job.
Dependent variable:

Job start month

 IV: Does loan size affect family formation?

Year/month of study completion FE
Age at completion spline
Additional controls

Marriage within 6 Parenthood within
years
6 years
-0.012 -0.006
-0.050
-0.044
(0.029) (0.025) (0.037) (0.035)
0.040***
0.032***
(0.004)
(0.004)
-0.018**
-0.024***
(0.008)
(0.009)
-0.084
0.001
(0.128)
(0.135)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
No
Yes

R-Squared
Percentage of successes
Observations

0.012
21.5
63,321

0.045
21.5
63,261

0.009
12.2
54,627

0.095
12.2
54,573

95% CI on coefficient on Loan value

-0.069
0.044

-0.055
0.044

-0.123
0.023

-0.112
0.024

Dependent variable:
Loan value at completion ($0,000s)
Female
NZ born
Pass rate (0-1)

Standard errors are robust. * p<0.10, ** p<0.05, *** p<0.01. Sample is students not already
married at completion (cols 1-2) or not already parents at completion (cols 3-4) whose first
destination was not overseas.

 Conclusions
• No cause for alarm that student loan size drives 
graduates overseas or prevents higher study 
• Some evidence that loan size matters for job strategy  
- graduates with larger loans may be more likely to 
migrate to a job and to secure a higher-paying job 
- suggests trade-off between earnings and lifestyle/
family considerations 
• Might expect graduates who benefit from Fees-Free 
policy to be more likely to take jobs where they want to 
live and to sacrifice some earnings for lifestyle factors