The institutions that directly impact economic growth are control of income inequality, economic openness, government expenditure, private investment, foreign direct investment. The feedback loops that contain these institutions have the most impact.

All of the feedbacks in the results are the same for all the countries. The difference from country to country depends on the projected trends in the future quality of the institutions.

We were surprised by the economic growth projected for Pakistan compared to the decline in economic growth in the other countries. We need to do more research into why this occurred.

We used the Euler integration function (default) to solve the differential equations in Vensim. And Euler integration can be translated into equations in the cells of Excel with each stock, flow, and auxiliary as a separate column in the spreadsheet. And in each row, the cells (stocks, flows, and auxiliaries) are updated for each step through time.

We wanted to select one country from each continent and secondly, we wanted to show that the model applied to countries with different levels of development. Also, we were particularly interested in countries facing different economic growth challenges.

1) The calibration of the model tries to draw a smooth line through the highly discontinuous datasets based on minimizing the total squared error. So the lines for some institutions will be a better match to the data than others. It may be worthwhile considering the addition of a few more variables.

2) Yes, we would be very open to the use of natural experiments. For example, many countries have attempted to increase their economic growth using various policies. It would be interesting to examine the results of these policies in natural experiments. This would be very useful to better understand the changes seen in the datasets for individual countries.

We used the data from 1996 to 2019. So the data from the financial crisis was included. However, some countries were more affected by the financial crisis than others. The model did not consider a shock to the system such as the 2008-9 financial crisis but in general System Dynamics can do that and often considers such shocks.

This work is a proof of concept at this time. So we used dimensionless units for policy effort. However, we recognize that policy implementation can be costly and will vary from institution to institution. So in the future, we would like to examine policy effort in terms of cost and effectiveness with the appropriate sensitivity analysis.

Based on econometric theory and results, there appears to be a strong cause-and-effect relationship between inflation and corruption in the literature. Practically, when there is increasing inflation (such as in Argentina and Turkey), there is a great loss in wealth and people may respond through corrupt behaviors to compensate.

When we change the effort, this affects the target level in the goal-gap formula of institutional change. So more effort is expended to achieve a higher target and the model responds by closing the gap between the target and the actual value over time

We did scenario analysis but not sensitivity analysis. Sensitivity analysis can be done quite easily in the model and it should be done in the future. We should have included a sensitivity analysis in the presentation.

They should have been labeled. Because we converted all of the values so that higher values were better, all of the loops were reinforcing loops and the differences in behavior occurred because the results of the reinforcing loops could lead to either exponential growth or exponential decline. And that demonstrates the importance of actually building a simulation model

Because all of the institutions are connected by reinforcing loops, that increasing economic growth will lead to increasing institutional quality and vice versa. We believe that increasing institutional quality is a good thing

Yes, the connection between government revenue and private investment needs to be looked at in more detail. This will be future work.

Yes, because all of the loops are reinforcing then the goals of the institutions are always to improve quality.

Yes, a country comparison is available but wasn’t included as part of the presentation. You can contact us to find out more.

Yes and no, this model is informing a member of the Planning and Development Board of Pakistan. However, the model has not been peer-reviewed. So there needs to be some caution in the interpretation and application of the results at this time. However, similar models have been developed and utilized by the World Bank and International Monetary Fund but not specifically System Dynamics models.

Yes, the model provides insights into all of the country analysis. In particular, the reinforcing loops in Pakistan result in growth while the reinforcing loops in the other countries are suggesting a decline. More work is necessary to examine these trends in more detail and determine if corrective action can be taken in the countries where decline is projected.

Yes, the institutions with closer connections to economic growth will have more impact on it. However, some of these will be synergistic and some will limit growth. So in fact, the leverage points are country-specific.

Yes, we have applied the model to the United States and Canada and it provided useful insights. We believe that model can be applied to any country in the world if there is data to support the calibration.

Yes, the change in institutional quality requires time and the times are calibrated in the model for all of the institutions separately. So some institutions will change faster and some will be delayed. Again, these times to change are country-specific.

At present, we haven’t looked at policy implementation. However, our work on policy optimization would require an estimation of the cost and benefits of policy options.

Yes, please contact us for more information in the case of the US or any other country.

Yes, income inequality, economic openness, government expenditure, private investment, and foreign direct investment had the most impact. Although this is true for all cases, the amount of impact varies from country to country.

Yes, please contact us and we will share these results.

No, we only tried to minimize the total sum of squared error for all of the institutions. However, it is possible to develop confidence intervals on the parameters and thereby, confidence intervals on the projections. This can be included in future work.