نعم، لا يوجد ما يمنع نظريًا من استخدام نسبة أو معدل (مثل نسبة التجارة إلى الناتج المحلي الإجمالي) كـ متغير تابع في نموذج ARDL، بشرط أن يكون لديك مبرر اقتصادي واضح لهذا الاختيار وأن يكون المتغير مناسبًا للغرض البحثي الخاص بك. نموذج ARDL يتطلب فقط وجود علاقة بين متغير تابع (يمكن أن يكون نسبة أو معدل) ومتغيرات تفسيرية أخرى

The answer depends on several factors. You may need to provide more details.

However, it is common for studies to use "ratios" as dependent variables taking into account the assumptions of the statistical tests being carried out.

It is also important to add that the use of the ARDL depends heavily on the results from the stationarity test.

The choice of the dependent variable is not dependent on the statistical method (like ARDL), but on the subject you want to analyse. If you want to explain the Trade/GDP ratio, then this ratio is a dependent variable. In ARDL, it (past values) is also an independent (explaining) variable.

Anton, my answer does belong to a different question. I have left the first point stand, as you have made a valuable comment about it.

I would be concerned about using a GDP deflator (in $ terms) to get a FDI series at constant $ prices. Perhaps you have no options.

John, are you sure that your answer belongs to this question?

By the way, I would never calculate or use a series of real FDIs. In fact, an FDI is a financial transaction, which, in general, adds nothing to productive capital (equipment). Therefore, it is not an investment in the sense of National accounts. If somebody buys a capital good (or shares), it is, of course, her individual investment, but at the same time it is a disinvestment of the seller. If both are from the same coutry, there will be no entry in economic statistics. If they are from different constituences, the transaction will show-up in both countries in BoP statistics with different signs. Nothing will (rightly) change in National Accounts (both on the supply and demand side, and also not what concerns a capital stock calculation).

John, I think, it is easy to construct a series of real FDIs. One could take the (nominal) FDIs in percent of nominal GDP and apply these percentages on real GDP. The problem is that I cannot imagine an analysis where these data can be sensefully used (applied) - especially for net FDIs.

Yes, using a ratio like Trade to GDP as a dependent variable in an ARDL (Autoregressive Distributed Lag) model is acceptable and often meaningful, especially when analyzing the relative importance or intensity of trade in relation to economic size; however, one must ensure the ratio is stationary or appropriately differenced, and its economic interpretation is well-grounded in theory.

Chuck, I doubt that it is meaningful to analyse the dependency of the Trade/GDP-ratio from country size with time-series analysis like ARDL. Logic says that this relation should be negative: e.g. for EU it is,of course, lower than for the EU member countries; when states separate (example: czechoslovakia), then Trade/GDP will increase. To find out empirical relations, I would never use time-series, but cross-section data for many (comparable) countries of different size.