It sounds like this same analysis generalizes to prove that increases in income tax rates do not reduce disposable income for the household sector (on the margin).
On another note, empirically the market seemed to react strongly positively to Trump corporate tax cuts. This suggests that either a) the market multiple/discount rate increased as a result of the change or b) expected future net income for the market increased. (Commentary at the time suggested b) was the main factor.) If we want to square this observation with your analysis, one explanation might be that net income rose for US-listed public companies at the expense of foreign-listed or domestic private companies. Or maybe it's a timing thing - the market expected that the necessary government spending cuts to finance the tax cut would come only after a lag, so corporate profits increase in the short run but not over the whole timeline.
No, it’s the market not understanding this analysis. Most investors in Wall Street think lower taxes increase profits. That is wrong and the market will find that out at some point. Until then enjoy the rally which is built on sand.
I'm not sure I follow the last paragraph. If companies are investing in profit-making ventures in a 0% tax scenario then definitionally they are making productive allocations of capital. That is why they are making a profit!
If your argument is that government somehow can more productively allocate capital even than companies using it for profit-making activities, then you are in fact in the end making an argument for communism, because what you are saying is tantamount to 'government can make more productive use of resources even than profitable companies'. If that is true in scenario X, then why not in scenarios Y and Z, and all other scenarios, for that matter?
There are poor allocations of capital at corporations and poor allocations in the government. As an investor, I can recognize poor (or good) in corporations a lot more effectively than government spending. There are certainty necessary government expenses covered by tax receipts. (how do you measure return on defense spending?) Thank you for an interesting post!
I repeated the above analysis between 1947-2024 for the top marginal tax rate and corporate profit/GDP ratio, and the correlation coefficient is significantly lower.
Fair enough, but what that tells you is that the extremely high corporate tax rates in the 1940s and 1950s have an impact. Since the 1960s, corporate tax rates have been much lower and the correlation disappears. As I say at the end of my article, there clearly is some link between tax rates and corporate profits for very high tax rates, but we are very, very far away from that point today.
Hmm, interesting. I checked my calculations and still come up with the same numbers that I mention in the article. No idea where the difference comes from.
Many apologies for not replying to you sooner, but I was travelling for business so could only do minimal maintenance on the conversations on this substack.
I looked into your data and compared it mine and I think I found the difference.
When I calculate the correlation between the top corporate tax rate and profit_gdp I get a similar correlation of -0.56. I use data since 1960 not 1947, so that might explain the small difference.
But I think the mistake in this approach is not to account for collinearity. The top marginal tax rate stays level for most of the time while profit_gdp also is relatively stable over time, so there is some collinearity there that might influence the correlation.
Having said that, the argument is that changes in corporate tax rates change profit_gdp, not that levels in corporate tax rates influence levels of profit_gdp.
If I take annual changes in top marginal tax rates and correlate them with changes in profit_gdp, I get a correlation of -0.16 for year-on-year changes. If I calculate the two-year changes in tax rates and correlate them with two-year changes in profit_gdp I get -0.07 and similarly -0.06 for rolling three-year changes.
Also, I delete all the periods where there is no change in corporate tax rates because that should not influence the correlation.
Thanks for the clarification. In the original analysis (tax rate and profit_gdp), the issue is that there is strong positive autocorrelation in the regression residuals (Durbin-Watson 0.43). This violates one of the key assumptions of OLS, the independence of errors, which distorts the correlation coefficient. When examining annual changes, I can also find negative correlation coefficients (consistent with what you observed), but the p-value in all cases is greater than 0.1. This indicates that there is no relationship between the two variables.
Well, the whole point of the argument is that there is no relationship between profits and tax rates…
The common conception is that higher tax rates ‘lead
To lower corporate profits’ so correlations should be negative and significantly different from zero.
Kalecky-Levy argues the correlation should be zero or not statistically significantly different from zero, which is what I find when I look at first differences, I.e. Changes in tax rates.
It sounds like this same analysis generalizes to prove that increases in income tax rates do not reduce disposable income for the household sector (on the margin).
On another note, empirically the market seemed to react strongly positively to Trump corporate tax cuts. This suggests that either a) the market multiple/discount rate increased as a result of the change or b) expected future net income for the market increased. (Commentary at the time suggested b) was the main factor.) If we want to square this observation with your analysis, one explanation might be that net income rose for US-listed public companies at the expense of foreign-listed or domestic private companies. Or maybe it's a timing thing - the market expected that the necessary government spending cuts to finance the tax cut would come only after a lag, so corporate profits increase in the short run but not over the whole timeline.
No, it’s the market not understanding this analysis. Most investors in Wall Street think lower taxes increase profits. That is wrong and the market will find that out at some point. Until then enjoy the rally which is built on sand.
I'm not sure I follow the last paragraph. If companies are investing in profit-making ventures in a 0% tax scenario then definitionally they are making productive allocations of capital. That is why they are making a profit!
If your argument is that government somehow can more productively allocate capital even than companies using it for profit-making activities, then you are in fact in the end making an argument for communism, because what you are saying is tantamount to 'government can make more productive use of resources even than profitable companies'. If that is true in scenario X, then why not in scenarios Y and Z, and all other scenarios, for that matter?
There are poor allocations of capital at corporations and poor allocations in the government. As an investor, I can recognize poor (or good) in corporations a lot more effectively than government spending. There are certainty necessary government expenses covered by tax receipts. (how do you measure return on defense spending?) Thank you for an interesting post!
I repeated the above analysis between 1947-2024 for the top marginal tax rate and corporate profit/GDP ratio, and the correlation coefficient is significantly lower.
corr(profit_gdp, taxrate) = -0,52925050
Under the null hypothesis of no correlation:
t(75) = -5,40204, with two-tailed p-value 0,0000
data source: https://tradingeconomics.com/united-states/corporate-tax-rate
https://fred.stlouisfed.org/graph/?g=1Pik
Fair enough, but what that tells you is that the extremely high corporate tax rates in the 1940s and 1950s have an impact. Since the 1960s, corporate tax rates have been much lower and the correlation disappears. As I say at the end of my article, there clearly is some link between tax rates and corporate profits for very high tax rates, but we are very, very far away from that point today.
I agree with you and the message of your article, but the relationship between the two variables is stronger.
From 2000 onwards, the correlation coefficient is -0.39 (p=0.09!!).
From 1990 onwards, the correlation coefficient is -0.42 (p=0.018).
From 1980 onwards, the correlation coefficient is -0.57 (p<0.001).
From 1960 onwards, the correlation coefficient is -0.53 (p<0.001).
Hmm, interesting. I checked my calculations and still come up with the same numbers that I mention in the article. No idea where the difference comes from.
Perhaps we used different data sources, which could explain the discrepancy. If you send me your data sources, I'd be happy to take a look.
Those are significant differences
Many apologies for not replying to you sooner, but I was travelling for business so could only do minimal maintenance on the conversations on this substack.
I looked into your data and compared it mine and I think I found the difference.
When I calculate the correlation between the top corporate tax rate and profit_gdp I get a similar correlation of -0.56. I use data since 1960 not 1947, so that might explain the small difference.
But I think the mistake in this approach is not to account for collinearity. The top marginal tax rate stays level for most of the time while profit_gdp also is relatively stable over time, so there is some collinearity there that might influence the correlation.
Having said that, the argument is that changes in corporate tax rates change profit_gdp, not that levels in corporate tax rates influence levels of profit_gdp.
If I take annual changes in top marginal tax rates and correlate them with changes in profit_gdp, I get a correlation of -0.16 for year-on-year changes. If I calculate the two-year changes in tax rates and correlate them with two-year changes in profit_gdp I get -0.07 and similarly -0.06 for rolling three-year changes.
Also, I delete all the periods where there is no change in corporate tax rates because that should not influence the correlation.
I hope that explains the difference.
Thanks for the clarification. In the original analysis (tax rate and profit_gdp), the issue is that there is strong positive autocorrelation in the regression residuals (Durbin-Watson 0.43). This violates one of the key assumptions of OLS, the independence of errors, which distorts the correlation coefficient. When examining annual changes, I can also find negative correlation coefficients (consistent with what you observed), but the p-value in all cases is greater than 0.1. This indicates that there is no relationship between the two variables.
Well, the whole point of the argument is that there is no relationship between profits and tax rates…
The common conception is that higher tax rates ‘lead
To lower corporate profits’ so correlations should be negative and significantly different from zero.
Kalecky-Levy argues the correlation should be zero or not statistically significantly different from zero, which is what I find when I look at first differences, I.e. Changes in tax rates.