Model Use Boundaries

Research-use boundaries for theoretical graphs, simulator outputs, empirical charts, forecasts, and policy interpretation.

Macro by Mark graphs can support teaching, research drafts, policy memos, and professional analysis. They do not remove the need to state the model, assumptions, data source, and limits of inference.

The main question is not whether a graph looks publication-ready. The question is what the graph is evidence for.

Theoretical Model Graphs

Theory-based model graphs show analytical structure: equilibrium, comparative statics, transition behavior, and scenario logic. They are appropriate for:

  • Teaching a model.
  • Explaining a theoretical mechanism.
  • Showing a comparative-static result.
  • Preparing a clean figure for a paper section that already states the model.

They are not empirical evidence by themselves. A theory graph does not estimate a coefficient, prove a policy worked, or validate a forecast.

When using a theory graph in a paper, cite Macro by Mark for the figure if exported from the site, and cite the underlying model literature for the theory.

Simulator and Reference-Model Graphs

Simulator graphs show model-generated paths under a specified structure and parameter set. These can be useful for:

  • Comparing scenarios.
  • Demonstrating dynamics.
  • Stress-testing intuition.
  • Teaching why a shock path may jump, kink, decay, overshoot, or converge.

They should be labeled as simulation output unless the route explicitly documents an estimated empirical run. A stylized simulation is not a measured fact.

Empirical Data Charts

Empirical charts are only as strong as their data provenance. For official or provider-backed data, cite the provider and the relevant methodology, release, or series identifier.

Macro by Mark can help package the figure and source metadata, but the data provider remains the statistical source.

Examples:

  • FRED chart: cite FRED or the underlying agency source.
  • BLS labor chart: cite BLS survey or release methodology.
  • World Bank cross-country chart: cite World Bank indicator metadata and methodology.
  • IMF chart: cite the relevant IMF dataset and release notes.

Forecast and Lab Outputs

Forecast and lab outputs are conditional estimates. They depend on:

  • Data window.
  • Transformations.
  • Model class.
  • Hyperparameters or prior choices.
  • Evaluation split.
  • Revision vintage.
  • Any manual scenario assumptions.

Use forecast and lab graphs as generated estimates, not as guarantees. The export may be suitable for a research appendix when paired with run metadata, data citations, and a clear description of the estimation design.

Policy Interpretation

A model graph can support policy reasoning, but it does not determine policy by itself. Policy analysis also needs institutional constraints, mandate language, political economy, distributional effects, uncertainty, and implementation lags.

For policy-facing work, add:

  • The policy instrument.
  • The transmission channels.
  • Timing and lags.
  • Institutional constraints.
  • Data series monitored.
  • Failure modes.
  • Historical episodes or primary sources.

Research-Use Labels

LabelMeaning
Figure sourceMacro by Mark generated the displayed figure
Model sourceThe underlying model comes from the cited economic literature
Data sourceThe observations come from the cited provider
Simulation outputValues are generated by the model under stated assumptions
Empirical estimateValues are estimated from data under a documented method
Adapted figureThe exported figure was changed after download

These labels should stay separate. A paper should not blur figure source, model source, and data source into one citation.

Minimum Standard for External Use

Before using a graph in a paper, memo, or deck, document:

  1. What the graph is meant to show.
  2. Whether it is theoretical, simulated, empirical, or forecast-generated.
  3. Which model, scenario, and variables are shown.
  4. Which data provider or model literature must be cited.
  5. Whether the exported figure was edited.
  6. Where the manifest and plotted data are stored.

If those six items are clear, a reader can treat the figure as a traceable artifact rather than a decorative chart.