Econometrics · Policy evaluation · Applied dissertations
50 panel data dissertation research questions grouped by model, from fixed effects and difference-in-differences to dynamic GMM, cointegration, synthetic control and staggered policy adoption.
50 Research topics
10 Model groups
10 Data sources
4 Universities cited
2019–26 Thesis years
By Siddharth Gupta · Statssy
Panel data lets you study the same countries, households, firms or regions across multiple periods. The modelling challenge is deciding whether the research question calls for fixed effects, dynamic models or a credible policy-evaluation design.
These 50 topics provide a data structure, suggested econometric model, public data source and related university thesis reference. Check the variable coverage and identification assumptions before choosing a topic.
Which panel model do you need? Start with your research design and data structure. Choosing among estimators also requires attention to assumptions, sample size and whether causal identification is credible.
How to use this list Identify the question type and find the relevant modelling approach. Pick two or three topics and open the proposed data source. Count the units, years and treatment groups; note missing values and irregular observation periods. Read the sample university thesis and examine its model diagnostics and identification strategy. Review the plan with your supervisor or book a free consultation . Jump to a model group 01. Pooled OLS and why panels matter 02. Fixed effects models 03. Random effects and the Hausman test 04. Difference-in-differences 05. Event study designs 06. Dynamic panels and GMM 07. Panel logit and probit 08. Panel unit roots and cointegration 09. Synthetic control 10. Staggered adoption and modern DiD Panel model group 01 / 10
Pooled OLS and why panels matter Start here to show why a panel model is needed. Compare a pooled regression with fixed effects and explain the gap.
TOPIC 01 Does pooled OLS overstate the link between trade and growth compared with fixed effects? Study design Country panel Panel model Pooled OLS compared with fixed effects Dataset World Bank WDI Open TOPIC 02 Do human capital and investment explain income levels across countries over time? Study design Country panel Panel model Pooled OLS with clustered errors Dataset Penn World Table Open TOPIC 03 Is the link between schooling and earnings smaller once each person is followed over time? Study design Household panel Panel model Pooled OLS compared with individual fixed effects Dataset PSID Free login TOPIC 04 Do state unemployment rates move with house prices across US states? Study design State panel Panel model Pooled OLS compared with state fixed effects Dataset FRED Open TOPIC 05 Do democracies have lower top income shares once fixed country traits are removed? Study design Country panel Panel model Pooled OLS compared with two-way fixed effects Dataset QoG Data + World Inequality Database Open Panel model group 02 / 10
Fixed effects models Fixed effects remove everything about a unit that does not change over time. They are the workhorse of panel dissertations.
TOPIC 06 Does education spending raise growth across Indian states? Study design State panel Panel model Two-way fixed effects Dataset RBI DBIE Open TOPIC 07 Does fiscal decentralisation help development in Sub-Saharan Africa? Study design Country panel Panel model Country and year fixed effects Dataset QoG Data + World Bank WDI Open TOPIC 08 Does losing a job hurt health in the years that follow? Study design Household panel Panel model Individual fixed effects Dataset PSID Free login TOPIC 09 Do cold winters raise the share of household spending on energy? Study design Household panel Panel model Household fixed effects with clustered errors Dataset Understanding Society Free login TOPIC 10 Do drought years lower agricultural output across countries? Study design Country panel Panel model Two-way fixed effects Dataset World Bank WDI Open Panel model group 03 / 10
Random effects and the Hausman test Random effects let you keep variables that do not change over time. The Hausman test tells you whether that is safe.
TOPIC 11 Should health spending and life expectancy be modelled with fixed or random effects? Study design Country panel Panel model Hausman test with both models Dataset World Bank WDI Open TOPIC 12 Does race shape wage growth once you follow workers over time? Study design Household panel Panel model Random effects with time-invariant variables Dataset PSID Free login TOPIC 13 Do regional traits explain differences in wellbeing across the UK? Study design Household panel Panel model Multilevel random effects Dataset Understanding Society Free login TOPIC 14 Do election cycles change public spending across countries? Study design Country panel Panel model Correlated random effects (Mundlak) Dataset QoG Data Open Related thesis: Essays on Election Bianca Appel Kranzler · Claremont Graduate University · 2026 · Related methodological example TOPIC 15 Does women’s labour force participation rise with GDP per head? Study design Country panel Panel model Hausman test with both models Dataset ILOSTAT Open Panel model group 04 / 10
Difference-in-differences Difference-in-differences compares treated and untreated units before and after a policy. It is the most common causal method in policy dissertations.
TOPIC 16 Did Medicaid expansion change employment in the states that expanded? Study design State panel Panel model Difference-in-differences Dataset FRED Open TOPIC 17 Did pay transparency laws narrow the gender pay gap? Study design Household panel Panel model Difference-in-differences Dataset PSID Free login TOPIC 18 Did state child tax credit expansions raise single mothers’ employment? Study design Household panel Panel model Difference-in-differences Dataset PSID Free login TOPIC 19 Did stricter lockdowns cut deaths more in some countries than others? Study design Country panel Panel model Continuous treatment difference-in-differences Dataset Our World in Data COVID-19 Open TOPIC 20 Did the 2018 US-China tariffs change spending in exposed states? Study design State panel Panel model Difference-in-differences Dataset FRED Open Panel model group 05 / 10
Event study designs Event studies show the effect year by year before and after a shock. They also let you test for trends before the policy began.
TOPIC 21 How did house prices change in counties hit by major wildfires or mudslides? Study design County panel Panel model Event study with leads and lags Dataset FRED Open TOPIC 22 Do state sales tax holidays shift retail sales in the months around them? Study design State panel Panel model Event study Dataset FRED Open TOPIC 23 How did bank deposits and credit respond month by month after demonetisation? Study design State panel Panel model Event study Dataset RBI DBIE Open TOPIC 24 Did case numbers fall in the weeks after vaccination rollouts began? Study design Country panel Panel model Event study with country fixed effects Dataset Our World in Data COVID-19 Open TOPIC 25 Did new ethanol plants raise county incomes in the years after opening? Study design County panel Panel model Event study Dataset FRED Open View the printable 50-topic guide
Panel model group 06 / 10
Dynamic panels and GMM Dynamic panels include last year’s value as a predictor. GMM estimators handle the bias this creates.
TOPIC 26 Does central bank independence lower how long inflation persists? Study design Country panel Panel model System GMM Dataset QoG Data + World Bank WDI Open TOPIC 27 Do poorer countries catch up once past growth is accounted for? Study design Country panel Panel model Difference GMM (Arellano-Bond) Dataset Penn World Table Open TOPIC 28 Do capital flows to emerging markets follow US monetary policy? Study design Country panel Panel model System GMM Dataset World Bank WDI + FRED Open TOPIC 29 Does credit growth feed on itself across countries? Study design Country panel Panel model System GMM Dataset World Bank WDI Open TOPIC 30 Does unemployment stay high long after a recession? Study design Country panel Panel model Dynamic panel with GMM Dataset ILOSTAT Open Panel model group 07 / 10
Panel logit and probit Panel logit and probit models suit yes or no outcomes measured over time, like buying a home or entering a crisis.
TOPIC 31 Who moves into homeownership, and when? Study design Household panel Panel model Fixed effects (conditional) logit Dataset PSID Free login TOPIC 32 What makes a household start saving for retirement? Study design Household panel Panel model Random effects probit Dataset PSID Free login TOPIC 33 Which countries enter IMF programmes? Study design Country panel Panel model Panel probit Dataset QoG Data + World Bank WDI Open TOPIC 34 Do banking crises follow credit booms? Study design Country panel Panel model Panel logit Dataset World Bank WDI Open TOPIC 35 Does unemployment raise the chance of reporting poor health? Study design Household panel Panel model Random effects logit Dataset Understanding Society Free login Panel model group 08 / 10
Panel unit roots and cointegration Long panels need unit root and cointegration tests first. Otherwise you risk finding links that are not real.
TOPIC 36 Do energy use and GDP move together in the long run across countries? Study design Country panel Panel model Pedroni cointegration test Dataset World Bank WDI Open TOPIC 37 Do savings and investment move together across countries? Study design Country panel Panel model Panel cointegration (Feldstein-Horioka test) Dataset Penn World Table Open TOPIC 38 Do house prices in US states share a long-run trend? Study design State panel Panel model IPS unit root test with Westerlund cointegration Dataset FRED Open TOPIC 39 Do health spending and GDP rise together in the long run? Study design Country panel Panel model Panel FMOLS Dataset World Bank WDI Open TOPIC 40 Do incomes in Indian states converge in the long run? Study design State panel Panel model Panel unit root tests on income gaps Dataset RBI DBIE Open Panel model group 09 / 10
Synthetic control Synthetic control builds a weighted twin from untreated units. It suits policies that hit one country or one state.
TOPIC 41 How would California’s housing supply look without its recent reforms? Study design State panel Panel model Synthetic control Dataset FRED Open TOPIC 42 Did Poland’s market reforms raise growth more than a synthetic Poland? Study design Country panel Panel model Synthetic control Dataset World Bank WDI Open TOPIC 43 Did dollarisation lower inflation compared with a synthetic peer? Study design Country panel Panel model Synthetic control Dataset World Bank WDI Open TOPIC 44 Did Sweden’s lighter COVID-19 rules cost more lives than a synthetic Sweden? Study design Country panel Panel model Synthetic control Dataset Our World in Data COVID-19 Open TOPIC 45 Did a large state tax cut change growth compared with a synthetic state? Study design State panel Panel model Synthetic control Dataset FRED Open Panel model group 10 / 10
Staggered adoption and modern DiD When states adopt a policy in different years, standard two-way fixed effects can mislead. New estimators fix this.
TOPIC 46 Did ban-the-box laws adopted in different years change employment? Study design State panel Panel model Callaway and Sant’Anna estimator Dataset PSID Free login TOPIC 47 Did universal pre-K policies rolled out over time raise enrolment? Study design District panel Panel model Staggered event study (Sun and Abraham) Dataset FRED Open TOPIC 48 Did state drug policy reforms adopted at different times change outcomes? Study design State panel Panel model Stacked difference-in-differences Dataset FRED Open TOPIC 49 Did farm loan waivers rolled out across Indian states change agricultural credit? Study design State panel Panel model Callaway and Sant’Anna estimator Dataset RBI DBIE Open TOPIC 50 Did state abortion restrictions change women’s economic outcomes? Study design State panel Panel model Two-way fixed effects with Goodman-Bacon decomposition Dataset PSID Free login View the printable 50-topic guide
10 datasets for panel-data econometrics The following sources support country, regional and household panels. Many proposed policy designs also need a separate treatment-date source or additional outcome variables.
Annual cross-country data covering growth, education, trade, health and energy. Check gaps and indicator-specific coverage.
Internationally comparable GDP, labour, capital and productivity data.
PSID Free login Long-running US household panel for within-person effects and longitudinal binary outcomes.
UK longitudinal household data, including work, expenditure and wellbeing; access varies by data tier.
Governance, corruption, institutions and political indicators at country-year level.
Indian state and banking data on government finances, credit and economic indicators.
US state and county economic indicators. Combine with separate policy adoption records for causal designs.
Historical COVID-19 data repository; check archival status and indicator availability.
Labour market measures by country, time, gender and age.
Income and wealth shares for inequality and economic distribution studies.
Printable econometrics dissertation guide
Get all 50 topics in one PDF The printable guide will contain the 50 research questions, panel structures, econometric models, dataset links and university thesis references. The download link will be added once the PDF is created and uploaded to WordPress.
PDF download coming soon How to turn one of these into your own topic Count units and years. Panel dimensions, missingness and persistence determine which models are feasible.Compare fixed and random effects carefully. Consider unobserved heterogeneity and whether regressors correlate with unit effects; the Hausman test is not a universal decision rule.Check DiD assumptions. Define treatment timing, inspect pre-treatment outcomes and consider alternative shocks and spillovers.Plan standard errors. Account for serial correlation, clustering and cross-sectional dependence as relevant.Use modern staggered DiD when needed. Consider cohort-time treatment-effect estimators rather than relying exclusively on conventional two-way fixed effects.For additional examples, see economics research topics , finance research topics and dissertation topics using regression analysis .
Related topic collections Need a second pair of eyes?
Choose a panel model your data supports. Bring your shortlist, time periods and sample structure to a free 15-minute discussion of estimators and identification assumptions.
Book a free 15-minute call 