Tape 85 - Computer-base economic forecasts

Download
- Rights
- Limited Re-Use
  CC BY-NC 4.0
  More Rights Info
- Files (1)
- MP3
- Please be patient with media downloads. They are often large files.
Share

Embed Code
<iframe src='//idn.duke.edu/ark:/87924/r4gf0nc53?embed=true' height='125px' width='100%' frameborder='0' marginwidth='0' marginheight='0' scrolling='no' allowfullscreen></iframe>
Permalink

https://idn.duke.edu/ark:/87924/r4gf0nc53
Skip to Item Info

Loading the media player...

Files for this Item

Transcript

Transcripts may contain inaccuracies.

-	Welcome once again, as MIT Professor, Paul Samuelson,	0:02
	discusses the current economic scene.	0:05
	This biweekly series is produced	0:07
	by Instructional Dynamics Incorporated	0:09
	and was recorded September 20, 1971.	0:11
-	I'd like today to comment on the new economic outlook	0:15
	as it seems to be affected by the Nixon economic program	0:20
	announced since August 15th.	0:25
	The first reaction to that program in terms of public	0:30
	opinion seems to have been extremely favorable.	0:33
	Gallup and Harris Polls show approval	0:36
	of something like 70% of the people,	0:41
	which is a very high number when you consider	0:45
	how many people are usually undecided,	0:47
	and how many people are divided on almost every issue.	0:49
	Similarly, the first impact of the announcement	0:55
	by the President on a Sunday evening, was to send	0:59
	the stock market up very substantially.	1:03
	By the time of the opening on Monday,	1:06
	it was too late to profit from the announcement	1:08
	and many stocks couldn't open all that day,	1:12
	but you will recall that the New York Stock Exchange	1:15
	had it's busiest day ever,	1:18
	in terms of total number of shares.	1:21
	I think that was just something in the 30 millions.	1:23
	It was a very profitable day for the brokerage fraternity	1:26
	and on paper,	1:29
	it was a very profitable day for the holders	1:31
	of common stocks	1:33
	because the market	1:35
	went up very considerably.	1:38
	I think something like 30 points, immediately.	1:40
	Indeed from an area in between 800 and 900,	1:44
	the market recovered as a result	1:49
	of the President's announcement,	1:52
	and went above 900.	1:53
	Likewise, the bond market did better.	1:58
	There was a present really for everybody,	2:01
	except those who had sold America short	2:03
	by being short in stocks.	2:05
	The bond markets went up	2:08
	for a couple of weeks	2:12
	and yields went down.	2:14
	This has been a fairly pervasive phenomenon.	2:18
	Of course, that kind of movement cannot last	2:22
	and there has been a period of reconsideration	2:27
	and consolidation in more recent weeks.	2:30
	Indeed, the market has come off it's peak	2:36
	and there has been some tendency for bonds to decline	2:39
	and for interest rates to harden.	2:44
	It still is true though,	2:48
	I believe that interest yields are less	2:49
	than before the President spoke.	2:51
	Now, was all this a rational reaction?	2:55
	Are these more favorable expectations merited?	3:01
	That's a question that we have to ask ourselves.	3:05
	Let the review then,	3:09
	some of the forecasts that are being made, currently,	3:11
	either by people in testimony	3:16
	before the Joint Economic Committee,	3:18
	as I speak those testifyings are not yet over,	3:20
	or in the public press	3:25
	or very many of the private forecasts.	3:28
	The quickest people off the mark, surprisingly,	3:32
	were some of the computer model builders.	3:35
	You'd of thought that the back of the envelope types	3:39
	who can change their mind in a flash	3:41
	would perhaps have been the first	3:44
	but they didn't have the confidence,	3:47
	which a computer gives you,	3:49
	And it was the people like Doctor Otto Eckstein	3:51
	of Data Resources Institute,	3:55
	who has a very popular data bank service,	3:58
	who is one of the first to publish new estimates	4:04
	In fact he testified before the Joint Economic Committee.	4:08
	His testimony must have been pleasing	4:12
	to the ears of most of his listeners	4:14
	because Doctor Eckstein first ran through a solution	4:17
	for the year 1972 based upon no new Nixon program,	4:22
	how things looked prior to the August 15th bombshell.	4:28
	And then using that as a control,	4:32
	he gave the new estimates.	4:35
	Now, I'm going to quote from memory and I think that,	4:37
	quantitatively, the impressions that I give you	4:41
	are going to be correct,	4:43
	but I don't have his testimony right before me.	4:44
	His fundamental diagnosis	4:48
	was extremely optimistic.	4:52
	He had real output growing	4:56
	in the year 1972 by seven percent,	4:58
	a good two percent more than the five percent number of his	5:02
	control solution prior to the President's new program.	5:07
	Was this substantial acceleration	5:12
	in the growth of real output,	5:14
	this more rapid return towards high employment,	5:16
	purchased at the expense of further inflation,	5:19
	not at all.	5:23
	Before the President's new program,	5:26
	Doctor Eckstein's computer estimated that prices,	5:28
	generally as measured I think by the GMPD Flator,	5:32
	would be growing in 1972 at about the rate 4.1% per annum.	5:35
	But with the President's program the wage price freeze	5:42
	and what Doctor Eckstein thought would follow it,	5:47
	my recollection is that the rate of price increase	5:51
	in 1972 as estimated by Doctor Eckstein,	5:54
	goes to below 3%,	5:57
	two and some fraction of a percent annual rate.	5:59
	Well, that is good news.	6:04
	It means that the high rates of unemployment	6:07
	which we've been having,	6:11
	would come down more rapidly as a result	6:13
	of the President's new program,	6:15
	than the control solution calculated	6:16
	prior to the new program, would've suggested.	6:20
	My recollection is that before the President's new program,	6:23
	Doctor Eckstein's computer model estimated	6:28
	by the end of the year, unemployment,	6:31
	this is by the end of 1972.	6:33
	The other numbers that I've been giving you	6:35
	have been comparisons,	6:36
	I suppose, of '72 and '71.	6:37
	But now, unemployment in the last month of the year,	6:39
	last quarter of the year,	6:43
	would be down to 5%	6:44
	instead of being 5.7% in the previous model.	6:48
	I think, in the previous model for the year,	6:52
	unemployment average is something like,	6:55
	I can't remember, but it's above 5 1/2%	6:58
	in the model, 5.9%, I think,	7:01
	and that's brought down that average number for the year	7:04
	to something like 5.3, 5.4 percent.	7:08
	Finally, just to complete the good news,	7:12
	Doctor Eckstein's profit estimates of profit	7:16
	after taxes show a substantial increase.	7:20
	I think it's from something like, 47 to 51 billion dollars.	7:25
	I'm not absolutely sure of that number,	7:31
	but something like a 10% increase in after tax profits.	7:33
-	Well, what are we to think?	7:41
	Let's reserve judgment for a moment	7:43
	and look at some of the other models.	7:44
	The Horton School model of Doctor Lawrence Klein also made	7:47
	a first pass at the President's new program and the results	7:53
	of it's calculations were equally optimistic,	7:59
	as I remember it.	8:04
	It later made a second pass and shaded the markups	8:05
	in the rate of real growth and the speeding up	8:10
	of the rate of unemployment reduction.	8:14
	Let me tell you a third model.	8:19
	The third model is that of the University of Michigan.	8:21
	The University of Michigan,	8:24
	the Ann Arbor group,	8:26
	made a pass at this problem and it seems to come up with	8:28
	the most optimistic unemployment estimates of all,	8:33
	because it has unemployment by the end of 1972,	8:36
	down in the 4 1/2% range and below.	8:41
	I will remind you that the last unemployment number	8:47
	that is knowable at the time that I'm speaking to you,	8:50
	is the August number,	8:54
	which was a blip upward to 6.1%.	8:56
	So the drop from 6.2% to 5.6% and then	8:59
	the rise to 5.8%, now to 6.1%,	9:03
	suggests that it's a safe statement to make,	9:07
	the only cautious statement that one can make,	9:11
	that unemployment has been hovering around the 6% mark	9:14
	for half a year now or more and that there is no, as yet,	9:17
	clear cut break out trend from that particular range.	9:22
	It would indeed be pleasant news	9:26
	that by the end of 1972,	9:30
	we would be, so nicely,	9:35
	back toward a 4% unemployment level.	9:38
-	What about government forecasts?	9:46
-	We've begun to have some testimony	9:51
	before a congressional committee,	9:53
	on the new program and I think I saw in the newspaper	9:55
	that Doctor George Shultz,	9:59
	the head of the Office of Budget of Management,	10:02
	has estimated that unemployment will be down	10:06
	to 5% or below	10:10
	by the middle of 1972.	10:13
	This is not quite so optimistic as the official estimate	10:17
	in the President's economic report given some months ago,	10:22
	where you'll recall,	10:27
	by the second quarter of 1972,	10:28
	unemployment was to be down to 4 1/2%.	10:31
	But even if we were down below 5% by then,	10:33
	that would be considered a fairly optimistic outcome	10:36
	and fairly satisfactory outcome.	10:42
	Forecasters however, have been impressed	10:45
	in this same optimistic way by the new program.	10:49
	Some Princeton forecasters, Ray Fair,	10:55
	I think perhaps helped by Professor William Branson,	11:00
	Professor Farron Branson, have made a forecast.	11:03
	And they show higher unemployment persisting	11:08
	than the numbers that we've just now been quoting.	11:13
	One of the more accurate forecasters whom I've been watching	11:17
	because it seems to me, he's had a very good track record,	11:22
	is Alan Greenspan of the Townsend-Greenspan Organization.	11:25
	Doctor Greenspan is not a government official	11:31
	but it's understood that he is an advisor of the President.	11:35
	He approves of the Nixon economic programs, generally,	11:41
	although he has some doubts,	11:48
	I think, about the new economic program,	11:50
	and he has made some write-ups in his rate of real growth.	11:53
	But he does not believe that just announcing a 90 day freeze	12:01
	and then following it by whatever	12:07
	bureaucratic structure is likely,	12:09
	really brings us out of the woods	12:14
	as far as the inflation problem is concerned.	12:15
	Actually, his computer forecast	12:19
	and his judgmental forecasts suggests that there will be	12:23
	a reacceleration of inflation as you move toward	12:26
	the last part of 1972 and into 1973.	12:32
	I believe he takes a dim view of putting	12:37
	too much steam in the boiler at this stage.	12:39
	He thinks that we will fritter away what advantage	12:42
	we may have gained by our past investment in a slowdown.	12:45
	So his particular numbers are much more modest.	12:50
	Now, let me comment on the methodology	12:55
	of some of these different forecasts.	12:57
	As far as one can tell,	13:02
	a computer does not enable the water	13:05
	to rise above it's own source.	13:08
	The computer is not more accurate than the judgments	13:11
	of the maker of the computer model.	13:15
	All that you can say about a computer model	13:20
	is that when it serves up to you a truth or a fiction,	13:22
	it will be a well-rounded plausible truth or fiction.	13:27
	The computer model as applied to GMP accounts	13:32
	tells a consistent story.	13:35
	A high computer forecast will tend to have all the different	13:39
	components of the vector of GMP high.	13:44
	They will add up to 100%.	13:47
	There will be consistency in their interrelations.	13:49
	And if that's the way nature's drawing from the box,	13:52
	the black box of the unknown future works out,	13:58
	that is on the high side,	14:01
	then the computer, generally,	14:02
	will tell you what things are going to look like.	14:04
	If on the other hand, nature,	14:07
	Mother Nature puts her hand into the black box	14:10
	of the uncertain future and draws out	14:12
	a low GMP overall forecast,	14:15
	then consistency requires that the various parts of the GMP,	14:19
	based upon past experience and past patterns,	14:26
	will also be low.	14:28
	The computer is a very fast way	14:30
	of telling a consistent story,	14:32
	but the story is told by a computer,	14:34
	alas, too often has to be a made up story.	14:37
	It has to be the gleam in the eye	14:41
	and the judgment of the computer maker.	14:44
	Well, so it is in this case.	14:48
	These first pass by the computer forecasts,	14:52
	which greatly write up the real GMP,	14:58
	and which considerably reduce the amount of unemployment	15:04
	as compared to a control solution,	15:07
	generally, proceed upon no more scientific basis	15:09
	than, let's assume say that the rate	15:15
	of increase in prices will be halved.	15:19
	Now, there's no accuracy until that particular judgment,	15:22
	no necessary accuracy, I should say.	15:28
	It may turn out to be very near the mark	15:30
	but it's not a judgment based	15:32
	upon elaborate regression equations	15:35
	and computer estimates of past experience.	15:37
	We don't have recent experience with a wage price freeze	15:41
	and of course nobody knows what,	15:44
	even the likelihoods are of the shape for the program	15:47
	that must necessarily follow after the 90 days.	15:51
	If you're going to make that kind of assumption,	15:57
	of course, you could've done that	15:58
	on the back of an envelope.	15:59
	If however, you done it on the back of an envelope,	16:01
	you would not immediately have to hand	16:03
	a consistent rendition of your fundamental,	16:05
	artistic perception in judgment.	16:09
	The result will look more scientific for the person	16:15
	who has used a computer	16:19
	but by a more fundamental	16:21
	understanding of what's involved in science	16:27
	and the difference between the science and art and judgment,	16:29
	there really won't be intrinsic difference in accuracy	16:34
	between the inconsistent forecast	16:37
	not done by a computer	16:41
	and the more consistent well-shaped,	16:42
	well-rounded forecast.	16:46
	It gives you a better fiction,	16:48
	if you want to put it that way.	16:49
	Another one of the models that I spoke about,	16:54
	which has not been tracking	16:57
	the behavior of inflation very well,	16:59
	was able to proceed upon the happy assumption that although,	17:02
	prior to Mr. Nixon's program,	17:07
	the regression equations of that model,	17:09
	based upon past experience were not tracking	17:11
	very accurately and without bias,	17:14
	the behavior of wages and prices.	17:16
	Indeed, they were systematically off.	17:19
	They were systematically making predictions	17:22
	of less inflation than actually materialized.	17:23
	So, they very happily put into their new computer run,	17:28
	the assumption that this time they're right	17:33
	and that they're previous regressions	17:35
	will now track correctly.	17:38
	Therefore, they didn't really	17:41
	have to change too much of their forecast.	17:42
	They just increased, I guess, their confidence	17:45
	in the correctness of that rather optimistic forecast.	17:48
	What I have to ask, is whether you and I can increase	17:53
	our confidence in those forecasts in the same way.	17:56
	I haven't discussed the monetarist forecasts	18:03
	but when I was looking over these different forecasts,	18:07
	I also looked over the forecast	18:10
	of the Federal Reserve Bank in St. Louis	18:13
	and of course, it was very little changed.	18:16
	In the first place,	18:20
	the Federal Reserve Bank of St. Louis assumes	18:21
	that the monetary magnitudes,	18:26
	the nominal value of the GMP,	18:30
	as against the real value of the GMP,	18:32
	is going to follow the same patterns with respect	18:34
	to the independent variables of that method of forecasting,	18:40
	as in the past and those independent variables of course,	18:42
	are primarily the rate of change	18:46
	of the money supply lagged over four quarters.	18:47
	Now, there's nothing in President Nixon's	18:51
	program of August 15th,	18:53
	which gives the Federal Reserve Bank of St. Louis	18:56
	any reason to change any of it's independent variables.	18:59
	Actually, of course,	19:04
	that particular monetarist model,	19:06
	like any monetarist model,	19:09
	is very cherry of doing any forecasting at all.	19:11
	It tells you, after the fact,	19:14
	where it thinks you have made mistakes in judgment.	19:16
	But there are very few monetarist's known to me	19:20
	who are ready to put their reputation's on the line	19:22
	and forecast what's going to be the future of the variables	19:27
	which they think are important.	19:30
	To explain the variables that the rest	19:32
	of us think important,	19:33
	I'm assuming that the rest of us are primarily concerned	19:35
	with the level of unemployment,	19:38
	with the level of real GMP,	19:40
	with the growth rates of that GMP,	19:42
	and with the rate of inflation.	19:45
	Now, take notes of our goals,	19:47
	and by the way, those are also,	19:51
	I don't mean to suggest otherwise,	19:53
	the goals of the monetarists,	19:55
	along on the behavior of those particular items.	19:57
	Let's see how by the monetarist method,	19:59
	you are able to predict	20:03
	what's going to happen to those goals,	20:06
	when you must predict the behavior	20:07
	of the monetary magnitudes.	20:10
	And President Nixon said nothing in his August 15th speech	20:12
	or in his subsequent speech to the Joint Houses of Congress,	20:17
	that would give anyone good leverage on changing	20:22
	his estimates with respect to the money supply.	20:27
	So what I did with the Federal Reserve Bank	20:29
	of St. Louis model was to put in the recent	20:31
	behavior of the monetary magnitude,	20:36
	so that's a matter of the past,	20:39
	and then make the most reasonable	20:41
	estimate that I would make,	20:43
	both in my own right and if I were a monetarist,	20:45
	for the behavior of the money supply in the rest	20:48
	of this year and going into next year.	20:51
	That, to me, suggested that in my choice between their 9%	20:55
	rate of growth of money model and their 6%, and their 3%,	21:01
	that there was no point in my looking at their 3% model.	21:04
	I followed their 9% model,	21:07
	even that was an understatement,	21:09
	for the recent period and then I	21:11
	modulated it into a 6% model.	21:13
	You know there's some evidence in the last few weeks,	21:17
	last month or more,	21:21
	that the rate of growth of the money supply	21:22
	as conventionally measured,	21:25
	has been subsiding,	21:26
	has been subsiding from say,	21:29
	an 11% rate toward a 5 or 6% rate.	21:31
	Same thing is to be noted in the M2 definition of money,	21:34
	including time deposits.	21:38
	Now, there are experts who think that this was practically,	21:42
	to be predicted because there's a bad	21:49
	seasonal correction in the rate of growth	21:51
	of the money supply in official figures.	21:53
	Professor Moti Aulani, my colleague,	21:57
	has pointed out to me that in either five	21:59
	out of the last six years,	22:02
	or was it six out of the last six years,	22:03
	there has been a seasonal movement of a more rapid rate	22:06
	of increase of the money supply in the first	22:10
	part of the year followed by a less	22:11
	rapid rate in the last part of the year.	22:13
	And so he's been fairly confident in predicting	22:16
	exactly what has happened would happen.	22:18
	You may ask, how is that possible	22:22
	if a seasonal correction is made properly,	22:23
	and my answer would be,	22:25
	that I suppose there not seasonally correcting	22:27
	the rate of growth of the money supply,	22:29
	which is what monetarists use,	22:30
	but their seasonally correcting the stock of money,	22:32
	not it's growth rate.	22:35
	And there could be no seasonal movement in one and still,	22:36
	I presume, a seasonal movement in the other.	22:40
	If I were a monetarist,	22:44
	I'd make my own seasonal adjustment	22:45
	based on getting rid of that particular phenomenon.	22:46
	But more than that,	22:49
	I think that the Federal Reserve has wanted to try harder	22:50
	to bring down the rates of growth of the money supply.	22:56
	We know that from the last Open Market Committee	22:58
	of something more than 90 days ago,	23:00
	which has had it's deliberations now made public.	23:03
	So most people had expected that this ought to happen	23:06
	and expected this would happen.	23:09
	It has happened.	23:10
	The Federal Reserve makes St. Louis	23:12
	in it's first pass at this,	23:14
	it's the most recent publication,	23:16
	was cherry of making any assumptions about changing	23:18
	the price behavior of the model and I suppose that it really	23:22
	invited it's own reader's to make those changes.	23:25
	Well now, where does that leave me?	23:28
	It leaves me with a feeling that I cannot believe	23:32
	in these simplistic computer models,	23:36
	which simply say, King Cnut has spoken,	23:39
	the president has said that prices and wages	23:42
	aren't to rise as rapidly as in the past.	23:43
	Therefore, it's as good as being done	23:46
	and we shade down our price increases.	23:48
	Moreover, it's bad methodology to assume	23:51
	that the money magnitudes remain unchanged	23:53
	and if you change your price assumption,	23:55
	then it all comes through to blow	23:57
	the line into the real assumption.	23:58
	Nevertheless, it does not leave me	24:00
	where I was before August 15th,	24:02
	because I think for better or worse,	24:04
	I think on the whole, for the better,	24:07
	the President has made a judgment	24:10
	that inflation is not as pressing	24:12
	a problem as is unemployment,	24:14
	in terms of politics,	24:17
	in terms of the welfare of the country,	24:18
	and he has relaxed his opposition to spending,	24:20
	he has told Congress that it can reduce the amount of taxes	24:24
	if Congress insists upon doing so,	24:27
	in departing from it's package.	24:30
	When I put all these things together,	24:31
	I come out with the conclusion that we're going to get	24:34
	some fiscal monetary expansion in real terms	24:38
	from this particular program.	24:41
	But I want to warn my listeners	24:43
	that 90 days lasts only 90 days	24:45
	and we're moving towards the end of that period.	24:48
	And there are gonna be some pretty thorny problems	24:51
	that have to be thrashed out in setting up the guide posts	24:55
	or whatever the modified system of wage	24:58
	and price control's will be after those 90 days.	25:01
-	If you have any questions or comments	25:04
	for Professor Samuelson,	25:06
	address them to Instructional Dynamics Incorporated,	25:07
	166 East Superior Street, Chicago, Illinois, 60611.	25:10

Item Info

Title:

Tape 85 - Computer-base economic forecasts

Date:

circa 1971

Contributor:

Samuelson, Paul A. (Paul Anthony), 1915-2009

Subject:

Format:

audiocassettes

Language:

English

Digital Collection:

Paul A. Samuelson papers, 1933-2010 and undated

Source Collection:

Paul A. Samuelson papers, 1933-2010

Series:

Economics Cassette Series: Paul Samuelson

Rights:

Limited Re-Use

CC BY-NC 4.0

Rights Note:

This material is made available for research, scholarship, and private study. Copyright in this material has not been transferred to Duke University. For reuses of this material beyond those permitted by fair use or otherwise allowed under the Copyright Act, please see our page on copyright and citations: https://library.duke.edu/rubenstein/research/citations-and-permissions.

Identifier:

RL10082-CS-0085
ref2309_3ta
samuelsonpaul
duke:325108
pstau001085
ark:/87924/r4gf0nc53
7f335162-4295-4ee4-ae8e-f752f5e5096b

Permalink:

https://idn.duke.edu/ark:/87924/r4gf0nc53

Sponsor:

Sponsor this Digital Collection

Tape 85 - Computer-base economic forecasts

Transcript

Item Info

Economists' Papers Archive

More Info & Options